[{"data":1,"prerenderedAt":2221},["ShallowReactive",2],{"blog-category-de-llmops":3},[4,78,125,182,251,313,370,440,494,548,606,660,706,813,903,1014,1084,1193,1252,1295,1378,1430,1483,1535,1614,1655,1703,1751,1799,1852,1907,1941,1989,2036,2079,2121],{"slug":5,"published":6,"minutes":7,"category":8,"tags":9,"tagPages":15,"keywords":17,"about":30,"sources":37,"cover":68,"og":69,"expertise":70,"locales":71,"lang":73,"title":75,"description":76,"coverAlt":77},"claude-plans-pricing-usage-limits","2026-10-21",11,"llmops",[10,11,12,13,14],"Claude","Pricing","Usage limits","Claude Code","LLM costs",[16],"claude-code",[18,19,20,21,22,23,24,25,26,27,28,29],"claude preise","claude pro vs max vergleich","claude max plan kosten","claude team preis","claude nutzungslimits","claude 5-stunden-limit","claude wochenlimit","claude code nutzungslimit","claude abo vs api","claude api-guthaben max-plan","claude kostenlos nutzen","claude tarife vergleich",[31,34],{"name":32,"url":33},"Claude (AI)","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FClaude_(AI)",{"name":35,"url":36},"Anthropic","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FAnthropic",[38,41,44,47,50,53,56,59,62,65],{"title":39,"url":40},"Anthropic pricing page (claude.com\u002Fpricing)","https:\u002F\u002Fclaude.com\u002Fpricing",{"title":42,"url":43},"Claude Help Center: What is the Pro plan?","https:\u002F\u002Fsupport.claude.com\u002Fen\u002Farticles\u002F8325606-what-is-the-pro-plan",{"title":45,"url":46},"Claude Help Center: What is the Max plan?","https:\u002F\u002Fsupport.claude.com\u002Fen\u002Farticles\u002F11049741-what-is-the-max-plan",{"title":48,"url":49},"Claude Help Center: What is the Team plan?","https:\u002F\u002Fsupport.claude.com\u002Fen\u002Farticles\u002F9266767-what-is-the-team-plan",{"title":51,"url":52},"Claude Help Center: Understanding usage and length limits","https:\u002F\u002Fsupport.claude.com\u002Fen\u002Farticles\u002F11647753-understanding-usage-and-length-limits",{"title":54,"url":55},"Claude Help Center: Usage limit best practices","https:\u002F\u002Fsupport.claude.com\u002Fen\u002Farticles\u002F9797557-usage-limit-best-practices",{"title":57,"url":58},"Claude Help Center: Models, usage, and limits in Claude Code","https:\u002F\u002Fsupport.claude.com\u002Fen\u002Farticles\u002F14552983-models-usage-and-limits-in-claude-code",{"title":60,"url":61},"Claude Platform docs: pricing","https:\u002F\u002Fplatform.claude.com\u002Fdocs\u002Fen\u002Fabout-claude\u002Fpricing",{"title":63,"url":64},"Anthropic: Introducing Claude Haiku 5.5 (7 October 2026)","https:\u002F\u002Fwww.anthropic.com\u002Fnews\u002Fclaude-haiku-5-5",{"title":66,"url":67},"Claude blog: Claude Cowork is now Claude","https:\u002F\u002Fclaude.com\u002Fblog\u002Fcowork-is-now-claude","\u002Fimages\u002Fblog\u002Fclaude-plans-pricing-usage-limits\u002Fcover.webp","\u002Fimages\u002Fblog\u002Fclaude-plans-pricing-usage-limits\u002Fog.jpg","ai-engineer",[72,73,74],"en","de","hu","Claude-Preise erklärt: Free, Pro, Max, Team und API sowie die Nutzungslimits","Claude-Tarife mit Preisen Stand Oktober 2026, Fünf-Stunden- und Wochenlimits, Team-Plätze, API-Preise pro Token und die Wahl des passenden Tarifs.","Claude-Preise: Drei unterschiedlich große Gläser voller Marken, daneben eine Uhr und eine Sanduhr.",{"slug":79,"published":80,"minutes":81,"category":8,"tags":82,"tagPages":88,"keywords":89,"about":99,"sources":102,"cover":119,"og":120,"expertise":70,"locales":121,"lang":73,"title":122,"description":123,"coverAlt":124},"claude-haiku-5-5","2026-10-19",15,[83,84,85,86,87],"Claude Haiku 5.5","LLM pricing","Reasoning effort","Model routing","Subagents",[],[90,91,79,92,93,94,95,96,97,98],"claude haiku 5.5","haiku 5.5 preise","haiku 5.5 vs haiku 4.5","haiku 5.5 vs sonnet 5.5 vergleich","haiku 5.5 benchmark ergebnisse","haiku 5.5 effort-stufen","claude subagent modell","günstigstes claude modell","haiku 5.5 api kosten",[100,101],{"name":32,"url":33},{"name":35,"url":36},[103,105,108,110,113,116],{"title":104,"url":64},"Anthropic: Introducing Claude Haiku 5.5 (October 7, 2026)",{"title":106,"url":107},"Claude Platform docs: Claude Haiku 5.5 overview","https:\u002F\u002Fplatform.claude.com\u002Fdocs\u002Fen\u002Fmodels\u002Fhaiku-5-5\u002Foverview",{"title":109,"url":61},"Claude Platform docs: Pricing",{"title":111,"url":112},"Claude Platform docs: Effort","https:\u002F\u002Fplatform.claude.com\u002Fdocs\u002Fen\u002Fbuild-with-claude\u002Feffort",{"title":114,"url":115},"Anthropic: Introducing Claude Sonnet 5.5 (September 28, 2026)","https:\u002F\u002Fwww.anthropic.com\u002Fnews\u002Fclaude-sonnet-5-5",{"title":117,"url":118},"Claude Platform docs: Models overview","https:\u002F\u002Fplatform.claude.com\u002Fdocs\u002Fen\u002Fabout-claude\u002Fmodels\u002Foverview","\u002Fimages\u002Fblog\u002Fclaude-haiku-5-5\u002Fcover.webp","\u002Fimages\u002Fblog\u002Fclaude-haiku-5-5\u002Fog.jpg",[72,73,74],"Claude Haiku 5.5: Preise, Effort-Stufen und welche Sonnet-Aufgaben wechseln können","Preise von Claude Haiku 5.5 je Preisstufe, Benchmarks gegen Haiku 4.5 und Sonnet 5.5, Effort-Stufen und welche Sonnet-5.5-Aufgaben wechseln können.","Claude Haiku 5.5: Ein leichter Papierflieger mit kleiner Münze zieht schnell an einem großen, schweren Gefäß vorbei.",{"slug":126,"published":127,"minutes":81,"category":8,"tags":128,"tagPages":134,"keywords":135,"about":149,"sources":157,"cover":176,"og":177,"expertise":70,"locales":178,"lang":73,"title":179,"description":180,"coverAlt":181},"deepseek-v4-1-flash-1017","2026-10-17",[129,130,131,132,133],"DeepSeek","Open-weight models","KV cache","Mixture of experts","LLM cost",[],[136,137,138,139,140,141,142,143,144,145,146,147,148],"deepseek v4.1 flash test","deepseek v4.1 flash kosten","kausaler encoder-decoder","compressed sparse attention 2 erklärt","csa2 cache-sharing","fp4-kv-cache","kv-cache komprimierung","swa bounded replay","engram conditional memory erklärt","speculative decoding deepseek","deepseek preise","deepswe benchmark ergebnisse","open-weight coding-modell",[150,152,154],{"name":129,"url":151},"https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FDeepSeek",{"name":132,"url":153},"https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FMixture_of_experts",{"name":155,"url":156},"Transformer (deep learning architecture)","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FTransformer_(deep_learning_architecture)",[158,161,164,167,170,173],{"title":159,"url":160},"DeepSeek-AI: DeepSeek-V4.1-Flash: Pushing the Limits of KV Cache Compression (arXiv:2609.19969)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2609.19969",{"title":162,"url":163},"Hugging Face: deepseek-ai\u002FDeepSeek-V4.1-Flash-1017 model card","https:\u002F\u002Fhuggingface.co\u002Fdeepseek-ai\u002FDeepSeek-V4.1-Flash-1017",{"title":165,"url":166},"OpenRouter: DeepSeek V4.1 Flash 1017, pricing and providers","https:\u002F\u002Fopenrouter.ai\u002Fdeepseek\u002Fdeepseek-v4.1-flash-1017",{"title":168,"url":169},"The Decoder: Deepseek V4.1 Flash 1017 beats larger rivals in agentic tasks while once again massively cutting per-task costs","https:\u002F\u002Fthe-decoder.com\u002Fdeepseek-v4-1-flash-1017-beats-larger-rivals-in-agentic-tasks-while-once-again-massively-cutting-per-task-costs\u002F",{"title":171,"url":172},"AI Weekly: DeepSeek opens V4.1-Flash-1017 weights and $0.75 Search API","https:\u002F\u002Faiweekly.co\u002Falerts\u002Fdeepseek-opens-v41-flash-1017-weights-and-075-search-api",{"title":174,"url":175},"DeepSeek on X: DeepSeek-V4.1-Flash-1017 announcement","https:\u002F\u002Fx.com\u002Fdeepseek_ai\u002Fstatus\u002F2111375640717033777","\u002Fimages\u002Fblog\u002Fdeepseek-v4-1-flash-1017\u002Fcover.webp","\u002Fimages\u002Fblog\u002Fdeepseek-v4-1-flash-1017\u002Fog.jpg",[72,73,74],"DeepSeek V4.1 Flash 1017: 890 Byte KV-Cache pro Token, und was ein Agent davon hat","DeepSeek V4.1 Flash 1017 im Deep Dive: Encoder-Decoder-Aufteilung, CSA2-Cache-Sharing, FP4-KV-Cache, Benchmarks und die echten Kosten pro Agenten-Task.","DeepSeek V4.1 Flash 1017: Ein Roboter mit winzigem Rucksack bricht neben einem riesigen Koffer zu einem langen Weg auf.",{"slug":183,"published":127,"minutes":184,"category":8,"tags":185,"tagPages":191,"keywords":192,"about":204,"sources":214,"cover":245,"og":246,"expertise":70,"locales":247,"lang":73,"title":248,"description":249,"coverAlt":250},"llm-rate-limits-retries-provider-fallback",10,[186,187,188,189,190],"Rate limits","Retries","Fallback","Reliability","LLM gateway",[],[193,194,195,196,197,198,199,200,201,202,203],"llm rate-limits behandeln","429 too many requests llm","exponentielles backoff mit jitter","anthropic 529 overloaded fehler","openai rate limit wiederholen","llm provider-fallback","litellm fallbacks einrichten","retry-after header erklärt","llm circuit breaker muster","llm-failover mehrere anbieter","claude api-limits",[205,208,211],{"name":206,"url":207},"Token bucket","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FToken_bucket",{"name":209,"url":210},"Exponential backoff","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FExponential_backoff",{"name":212,"url":213},"Circuit breaker design pattern","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FCircuit_breaker_design_pattern",[215,218,221,224,227,230,233,236,239,242],{"title":216,"url":217},"Claude Platform Docs: Rate limits","https:\u002F\u002Fplatform.claude.com\u002Fdocs\u002Fen\u002Fapi\u002Frate-limits",{"title":219,"url":220},"Claude Platform Docs: API errors","https:\u002F\u002Fplatform.claude.com\u002Fdocs\u002Fen\u002Fapi\u002Ferrors",{"title":222,"url":223},"OpenAI API Docs: Rate limits","https:\u002F\u002Fdevelopers.openai.com\u002Fapi\u002Fdocs\u002Fguides\u002Frate-limits",{"title":225,"url":226},"AWS Builders Library: Timeouts, retries and backoff with jitter","https:\u002F\u002Fbuilder.aws.com\u002Fcontent\u002F3EumjoZascWd1oZiEgL8ORlv3qE\u002Ftimeouts-retries-and-backoff-with-jitter",{"title":228,"url":229},"MDN: Retry-After header","https:\u002F\u002Fdeveloper.mozilla.org\u002Fen-US\u002Fdocs\u002FWeb\u002FHTTP\u002FReference\u002FHeaders\u002FRetry-After",{"title":231,"url":232},"Martin Fowler: Circuit Breaker (2014)","https:\u002F\u002Fmartinfowler.com\u002Fbliki\u002FCircuitBreaker.html",{"title":234,"url":235},"LiteLLM Docs: Router, retries and cooldowns","https:\u002F\u002Fdocs.litellm.ai\u002Fdocs\u002Frouting",{"title":237,"url":238},"LiteLLM Docs: Proxy fallbacks and reliability","https:\u002F\u002Fdocs.litellm.ai\u002Fdocs\u002Fproxy\u002Freliability",{"title":240,"url":241},"OpenRouter Docs: Model fallbacks","https:\u002F\u002Fopenrouter.ai\u002Fdocs\u002Fguides\u002Frouting\u002Fmodel-fallbacks",{"title":243,"url":244},"AI SDK Docs: Settings (maxRetries, timeout)","https:\u002F\u002Fai-sdk.dev\u002Fdocs\u002Fai-sdk-core\u002Fsettings","\u002Fimages\u002Fblog\u002Fllm-rate-limits-retries-provider-fallback\u002Fcover.webp","\u002Fimages\u002Fblog\u002Fllm-rate-limits-retries-provider-fallback\u002Fog.jpg",[72,73,74],"LLM-Rate-Limits in Produktion: Retries, Backoff und Provider-Fallback","LLM-Rate-Limits in Produktion: welche Fehler man wiederholt, Backoff mit Jitter, Retries nur auf einer Schicht und Provider-Fallback ohne kaputte Ausgaben.","LLM-Rate-Limits: Ein Lieferwagen mit Sanduhr wartet an einer Schranke, eine Nebenstraße führt zu einem offenen Tor.",{"slug":252,"published":127,"minutes":253,"category":8,"tags":254,"tagPages":260,"keywords":261,"about":272,"sources":282,"cover":307,"og":308,"expertise":70,"locales":309,"lang":73,"title":310,"description":311,"coverAlt":312},"prompt-versioning-management",8,[255,256,257,258,259],"Prompt management","Versioning","LLMOps","Evals","Langfuse",[],[262,263,264,265,266,267,268,269,270,271],"prompt-versionierung produktion","prompt management tool","prompt-registry","versionskontrolle für prompts","langfuse prompt-management","mlflow prompt registry einrichten","prompts in git speichern","openai prompt-objekt abkündigung","prompt rollback produktion","prompt-evaluation in ci",[273,276,279],{"name":274,"url":275},"Prompt engineering","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FPrompt_engineering",{"name":277,"url":278},"Version control","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FVersion_control",{"name":280,"url":281},"Continuous integration","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FContinuous_integration",[283,286,289,292,295,298,301,304],{"title":284,"url":285},"OpenAI API Docs: Migrate from prompt objects","https:\u002F\u002Fdevelopers.openai.com\u002Fapi\u002Fdocs\u002Fguides\u002Fprompting\u002Fmigrate-from-prompt-object",{"title":287,"url":288},"OpenAI API Docs: Prompt engineering (reusable prompts deprecation note)","https:\u002F\u002Fdevelopers.openai.com\u002Fapi\u002Fdocs\u002Fguides\u002Fprompt-engineering",{"title":290,"url":291},"Langfuse Docs: Prompt version control","https:\u002F\u002Flangfuse.com\u002Fdocs\u002Fprompt-management\u002Ffeatures\u002Fprompt-version-control",{"title":293,"url":294},"Langfuse Docs: Prompt caching in the SDKs","https:\u002F\u002Flangfuse.com\u002Fdocs\u002Fprompt-management\u002Ffeatures\u002Fcaching",{"title":296,"url":297},"Langfuse Docs: Link prompts to traces","https:\u002F\u002Flangfuse.com\u002Fdocs\u002Fprompt-management\u002Ffeatures\u002Flink-to-traces",{"title":299,"url":300},"MLflow Docs: Prompt Registry","https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Fprompt-registry\u002F",{"title":302,"url":303},"Promptfoo Docs: GitHub Action","https:\u002F\u002Fwww.promptfoo.dev\u002Fdocs\u002Fintegrations\u002Fgithub-action\u002F",{"title":305,"url":306},"Claude Platform Docs: Prompt caching","https:\u002F\u002Fplatform.claude.com\u002Fdocs\u002Fen\u002Fbuild-with-claude\u002Fprompt-caching","\u002Fimages\u002Fblog\u002Fprompt-versioning-management\u002Fcover.webp","\u002Fimages\u002Fblog\u002Fprompt-versioning-management\u002Fog.jpg",[72,73,74],"Prompt-Versionierung in Produktion: Git, Registries, Labels und Rollbacks","Prompts in Produktion versionieren: was zu einer Version gehört, Git oder Prompt-Registry, Labels und Rollbacks, Eval-Gates in CI und Folgen für Caches.","Prompt-Versionierung: Eine Hand zieht ein älteres Glas mit Schriftrolle aus einer beschrifteten Reihe.",{"slug":314,"published":127,"minutes":315,"category":8,"tags":316,"tagPages":321,"keywords":322,"about":333,"sources":343,"cover":364,"og":365,"expertise":70,"locales":366,"lang":73,"title":367,"description":368,"coverAlt":369},"llm-batch-api-offline-workloads",9,[317,318,257,319,320],"Batch API","Cost","Prompt caching","Data pipelines",[],[323,324,325,326,327,328,329,330,331,332],"llm batch-api","anthropic message batches api nutzen","openai batch api kosten","gemini batch-modus","batch-inferenz llm kosten","batch api 50 prozent rabatt","openai flex processing","bedrock batch-inferenz","llm offline-verarbeitung","batch api prompt caching kombinieren",[334,337,340],{"name":335,"url":336},"Batch processing","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FBatch_processing",{"name":338,"url":339},"JSON Lines","https:\u002F\u002Fjsonlines.org\u002F",{"name":341,"url":342},"Large language model","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FLarge_language_model",[344,347,349,350,353,356,358,361],{"title":345,"url":346},"Claude Platform Docs: Batch processing","https:\u002F\u002Fplatform.claude.com\u002Fdocs\u002Fen\u002Fbuild-with-claude\u002Fbatch-processing",{"title":348,"url":217},"Claude Platform Docs: Rate limits (Message Batches API)",{"title":305,"url":306},{"title":351,"url":352},"OpenAI API Docs: Batch API","https:\u002F\u002Fdevelopers.openai.com\u002Fapi\u002Fdocs\u002Fguides\u002Fbatch",{"title":354,"url":355},"OpenAI API Docs: Flex processing","https:\u002F\u002Fdevelopers.openai.com\u002Fapi\u002Fdocs\u002Fguides\u002Fflex-processing",{"title":357,"url":223},"OpenAI API Docs: Rate limits (batch queue limits)",{"title":359,"url":360},"Google AI for Developers: Gemini Batch API","https:\u002F\u002Fai.google.dev\u002Fgemini-api\u002Fdocs\u002Fbatch-api",{"title":362,"url":363},"Amazon Bedrock User Guide: Batch inference","https:\u002F\u002Fdocs.aws.amazon.com\u002Fbedrock\u002Flatest\u002Fuserguide\u002Fbatch-inference.html","\u002Fimages\u002Fblog\u002Fllm-batch-api-offline-workloads\u002Fcover.webp","\u002Fimages\u002Fblog\u002Fllm-batch-api-offline-workloads\u002Fog.jpg",[72,73,74],"LLM-Batch-APIs: Inferenz zum halben Preis für Arbeit, die warten kann","Wann sich die Batch-APIs von Anthropic, OpenAI und Gemini lohnen, wie Limits, Zuordnung und Fehler funktionieren und wie sich Rabatte mit Caching addieren.","LLM-Batch-APIs: Ein Postkorb voller Umschläge wartet über Nacht, daneben ein Preisschild mit halber Münze.",{"slug":371,"published":127,"minutes":7,"category":8,"tags":372,"tagPages":376,"keywords":377,"about":389,"sources":397,"cover":434,"og":435,"expertise":70,"locales":436,"lang":73,"title":437,"description":438,"coverAlt":439},"small-language-models-when-enough",[373,374,375,86,257],"Small language models","Distillation","Quantization",[],[378,379,380,381,382,383,384,385,386,387,388],"kleine sprachmodelle","wann ein kleines sprachmodell nutzen","slm vs llm kosten","modell-destillation","quantization-aware training","q4_k_m quantisierung","llm-routing kleine modelle","kleines modell evaluieren","on-device llm auf dem gerät","gemma 4 e4b modell","lokales llm datenschutz",[390,393,396],{"name":391,"url":392},"Small language model","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FSmall_language_model",{"name":394,"url":395},"Knowledge distillation","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FKnowledge_distillation",{"name":341,"url":342},[398,401,404,407,410,413,416,419,422,425,428,431],{"title":399,"url":400},"Belcak et al.: Small Language Models are the Future of Agentic AI (arXiv 2506.02153)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2506.02153",{"title":402,"url":403},"Google: Gemma 4 model card","https:\u002F\u002Fai.google.dev\u002Fgemma\u002Fdocs\u002Fcore\u002Fmodel_card_4",{"title":405,"url":406},"Google: Gemma 3 QAT models, state-of-the-art AI on consumer GPUs","https:\u002F\u002Fdevelopers.googleblog.com\u002Fen\u002Fgemma-3-quantized-aware-trained-state-of-the-art-ai-to-consumer-gpus\u002F",{"title":408,"url":409},"Google: Introducing Gemma 3 270M","https:\u002F\u002Fdevelopers.googleblog.com\u002Fen\u002Fintroducing-gemma-3-270m\u002F",{"title":411,"url":412},"Microsoft Research: Phi-4 Technical Report (arXiv 2412.08905)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2412.08905",{"title":414,"url":415},"Qwen: Qwen3 Technical Report (arXiv 2505.09388)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2505.09388",{"title":417,"url":418},"Apple: Apple Intelligence Foundation Language Models, Tech Report 2025 (arXiv 2507.13575)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2507.13575",{"title":420,"url":421},"Ong et al.: RouteLLM, Learning to Route LLMs with Preference Data (arXiv 2406.18665)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2406.18665",{"title":423,"url":424},"LMSYS: RouteLLM, an open-source framework for cost-effective LLM routing","https:\u002F\u002Flmsys.org\u002Fblog\u002F2024-07-01-routellm\u002F",{"title":426,"url":427},"Gomes et al.: On the Generalization vs Fidelity Paradox in Knowledge Distillation (arXiv 2505.15442)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2505.15442",{"title":429,"url":430},"llama.cpp: quantize tool documentation","https:\u002F\u002Fgithub.com\u002Fggml-org\u002Fllama.cpp\u002Fblob\u002Fmaster\u002Ftools\u002Fquantize\u002FREADME.md",{"title":432,"url":433},"European Commission: AI Act","https:\u002F\u002Fdigital-strategy.ec.europa.eu\u002Fen\u002Fpolicies\u002Fregulatory-framework-ai","\u002Fimages\u002Fblog\u002Fsmall-language-models-when-enough\u002Fcover.webp","\u002Fimages\u002Fblog\u002Fsmall-language-models-when-enough\u002Fog.jpg",[72,73,74],"Kleine Sprachmodelle: wann ein 4B-Modell reicht","Wann ein kleines Modell die Frontier-API schlägt: Kosten pro Fall, Latenz, Datenschutz, Destillation, Quantisierung, Routing und ein Eval-Plan.","Kleine Sprachmodelle: Eine Hand nimmt statt des großen Werkzeugkastens ein Taschenmesser, um eine Schnur zu schneiden.",{"slug":441,"published":127,"minutes":184,"category":8,"tags":442,"tagPages":445,"keywords":446,"about":458,"sources":464,"cover":488,"og":489,"expertise":70,"locales":490,"lang":73,"title":491,"description":492,"coverAlt":493},"reasoning-models-effort-thinking-budget",[443,444,257,318,258],"Reasoning models","Effort",[],[447,448,449,450,451,452,453,454,455,456,457],"reasoning effort parameter erklärt","claude effort einstellung","claude adaptive thinking aktivieren","budget_tokens abkündigung","openai reasoning effort stufen","gemini thinking_level einstellung","thinking-tokens kosten","reasoning-tokens abrechnung","claude task budget","overthinking reasoning-modelle","max_tokens thinking einstellen",[459,462,463],{"name":460,"url":461},"Reasoning language model","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FReasoning_language_model",{"name":341,"url":342},{"name":274,"url":275},[465,467,470,473,476,479,482,485],{"title":466,"url":112},"Claude Platform Docs: Effort",{"title":468,"url":469},"Claude Platform Docs: Extended thinking (manual budget_tokens mode)","https:\u002F\u002Fplatform.claude.com\u002Fdocs\u002Fen\u002Fbuild-with-claude\u002Fextended-thinking",{"title":471,"url":472},"Claude Platform Docs: Steering thinking (cost, caching, pricing)","https:\u002F\u002Fplatform.claude.com\u002Fdocs\u002Fen\u002Fbuild-with-claude\u002Fthinking-steering-and-cost",{"title":474,"url":475},"Claude Platform Docs: Task budgets","https:\u002F\u002Fplatform.claude.com\u002Fdocs\u002Fen\u002Fbuild-with-claude\u002Ftask-budgets",{"title":477,"url":478},"OpenAI API Docs: Reasoning models","https:\u002F\u002Fdevelopers.openai.com\u002Fapi\u002Fdocs\u002Fguides\u002Freasoning",{"title":480,"url":481},"Google AI for Developers: Gemini thinking","https:\u002F\u002Fai.google.dev\u002Fgemini-api\u002Fdocs\u002Fthinking",{"title":483,"url":484},"Chen et al.: Do NOT Think That Much for 2+3=? On the Overthinking of o1-Like LLMs (arXiv 2412.21187)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2412.21187",{"title":486,"url":487},"Gema et al.: Inverse Scaling in Test-Time Compute (arXiv 2507.14417)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2507.14417","\u002Fimages\u002Fblog\u002Freasoning-models-effort-thinking-budget\u002Fcover.webp","\u002Fimages\u002Fblog\u002Freasoning-models-effort-thinking-budget\u002Fog.jpg",[72,73,74],"Reasoning Effort und Thinking-Budgets: Wie viel soll ein Modell denken?","Wie Effort, adaptives Thinking und Thinking-Level bei Claude, OpenAI und Gemini funktionieren, was Thinking-Tokens kosten und wie ein Effort-Sweep hilft.","Reasoning Effort: Ein Roboter denkt unter einer Denkblase neben einer Sanduhr und einem Drehregler.",{"slug":495,"published":496,"minutes":315,"category":8,"tags":497,"tagPages":503,"keywords":504,"about":516,"sources":523,"cover":542,"og":543,"expertise":70,"locales":544,"lang":73,"title":545,"description":546,"coverAlt":547},"structured-outputs-json-schema-llm","2026-10-16",[498,499,500,501,502],"Structured outputs","JSON Schema","Constrained decoding","LLM APIs","Validation",[],[505,506,507,508,509,510,511,512,513,514,515],"structured outputs llm anleitung","json schema llm-ausgabe","openai structured outputs strict-modus","claude structured outputs einrichten","gemini structured output nutzen","constrained decoding erklärt","json mode vs structured outputs vergleich","llm json-ausgabe validieren","zod structured output ki","pydantic llm-ausgabe","vllm structured outputs einrichten",[517,520,522],{"name":518,"url":519},"JSON","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FJSON",{"name":499,"url":521},"https:\u002F\u002Fjson-schema.org\u002F",{"name":341,"url":342},[524,527,530,533,536,539],{"title":525,"url":526},"OpenAI API docs: Structured model outputs","https:\u002F\u002Fplatform.openai.com\u002Fdocs\u002Fguides\u002Fstructured-outputs",{"title":528,"url":529},"Claude API docs: Structured outputs","https:\u002F\u002Fplatform.claude.com\u002Fdocs\u002Fen\u002Fbuild-with-claude\u002Fstructured-outputs",{"title":531,"url":532},"Gemini API docs: Structured outputs","https:\u002F\u002Fai.google.dev\u002Fgemini-api\u002Fdocs\u002Fstructured-output",{"title":534,"url":535},"vLLM docs: Structured Outputs","https:\u002F\u002Fdocs.vllm.ai\u002Fen\u002Flatest\u002Ffeatures\u002Fstructured_outputs.html",{"title":537,"url":538},"Tam et al.: Let Me Speak Freely? A Study on the Impact of Format Restrictions on Performance of LLMs (2024)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2408.02442",{"title":540,"url":541},"Geng et al.: JSONSchemaBench: A Rigorous Benchmark of Structured Outputs for Language Models (2025)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2501.10868","\u002Fimages\u002Fblog\u002Fstructured-outputs-json-schema-llm\u002Fcover.webp","\u002Fimages\u002Fblog\u002Fstructured-outputs-json-schema-llm\u002Fog.jpg",[72,73,74],"Structured Outputs mit JSON Schema: was die Garantie abdeckt und was nicht","Wie Structured Outputs und Constrained Decoding bei OpenAI, Claude, Gemini und vLLM funktionieren, welche Schemaregeln zählen und was du validierst.","Structured Outputs: Ein Ausstecher formt einen perfekten Keks, daneben ein richtig geformter, aber rissiger.",{"slug":549,"published":496,"minutes":315,"category":8,"tags":550,"tagPages":556,"keywords":557,"about":570,"sources":578,"cover":600,"og":601,"expertise":70,"locales":602,"lang":73,"title":603,"description":604,"coverAlt":605},"evaluating-ai-agents-trajectories",[551,552,553,554,555],"Agent evals","Trajectories","Tool calling","pass^k","LLM-as-judge",[],[558,559,560,561,562,563,564,565,566,567,568,569],"ki-agenten evaluieren","trajektorien-evaluation agenten","ki-agenten evals","tool-call genauigkeit","trajektorien-vergleich evaluator","pass^k metrik","tau-bench","llm als richter für agenten","agent regressionstests","ergebnis- vs trajektorien-evaluation","agentevals","tool correctness metrik",[571,574,577],{"name":572,"url":573},"Intelligent agent","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FIntelligent_agent",{"name":575,"url":576},"Software testing","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FSoftware_testing",{"name":341,"url":342},[579,582,585,588,591,594,597],{"title":580,"url":581},"Anthropic: Demystifying evals for AI agents (Jan 2026)","https:\u002F\u002Fwww.anthropic.com\u002Fengineering\u002Fdemystifying-evals-for-ai-agents",{"title":583,"url":584},"Yao et al.: τ-bench: A Benchmark for Tool-Agent-User Interaction in Real-World Domains (2024)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2406.12045",{"title":586,"url":587},"Barres et al.: τ²-Bench: Evaluating Conversational Agents in a Dual-Control Environment (2025)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2506.07982",{"title":589,"url":590},"LangChain: agentevals (GitHub)","https:\u002F\u002Fgithub.com\u002Flangchain-ai\u002Fagentevals",{"title":592,"url":593},"DeepEval docs: Tool Correctness","https:\u002F\u002Fdeepeval.com\u002Fdocs\u002Fmetrics-tool-correctness",{"title":595,"url":596},"Ragas docs: Agentic or tool use metrics","https:\u002F\u002Fdocs.ragas.io\u002Fen\u002Fstable\u002Fconcepts\u002Fmetrics\u002Favailable_metrics\u002Fagents\u002F",{"title":598,"url":599},"OpenAI API docs: Trace grading","https:\u002F\u002Fplatform.openai.com\u002Fdocs\u002Fguides\u002Ftrace-grading","\u002Fimages\u002Fblog\u002Fevaluating-ai-agents-trajectories\u002Fcover.webp","\u002Fimages\u002Fblog\u002Fevaluating-ai-agents-trajectories\u002Fog.jpg",[72,73,74],"KI-Agenten evaluieren: Trajektorien, Ergebnisse und Zuverlässigkeit bewerten","KI-Agenten evaluieren: Ergebnis- und Trajektorie-Grader, Match-Modi, pass^k, isolierte Umgebungen und LLM-Judges, die belastbar sind.","Agenten-Evaluation: Die Fußspur eines Roboters durch ein Labyrinth wird von einem großen Bleistift geprüft.",{"slug":607,"published":496,"minutes":184,"category":8,"tags":608,"tagPages":614,"keywords":615,"about":627,"sources":634,"cover":654,"og":655,"expertise":70,"locales":656,"lang":73,"title":657,"description":658,"coverAlt":659},"llm-model-migration-regression-testing",[609,610,611,612,613],"Model migration","Regression testing","Deprecations","LLM evals","Canary releases",[],[616,617,618,619,620,621,622,623,624,625,626],"llm-modellmigration","llm-regressionstests","openai modell abschaltung","claude modell abkündigung","llm-modell wechseln ohne fehler","llm modell upgrade checkliste","prompt-regressionstests","schatten-tests llm","modell-abschaltdatum","golden dataset llm-evals","llm drift überwachen",[628,630,631],{"name":610,"url":629},"https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FRegression_testing",{"name":341,"url":342},{"name":632,"url":633},"Feature toggle","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FFeature_toggle",[635,638,641,644,647,650,653],{"title":636,"url":637},"Claude API docs: Model deprecations","https:\u002F\u002Fplatform.claude.com\u002Fdocs\u002Fen\u002Fabout-claude\u002Fmodel-deprecations",{"title":639,"url":640},"OpenAI API docs: Deprecations","https:\u002F\u002Fplatform.openai.com\u002Fdocs\u002Fdeprecations",{"title":642,"url":643},"Gemini API docs: Models and version name patterns","https:\u002F\u002Fai.google.dev\u002Fgemini-api\u002Fdocs\u002Fmodels",{"title":645,"url":646},"Claude API docs: Claude Opus 5.5 migration guide","https:\u002F\u002Fplatform.claude.com\u002Fdocs\u002Fen\u002Fmodels\u002Fopus-5-5\u002Fmigration-guide",{"title":648,"url":649},"Chen, Zaharia, Zou: How is ChatGPT's behavior changing over time? (2023)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2307.09009",{"title":651,"url":652},"Sclar et al.: Quantifying Language Models' Sensitivity to Spurious Features in Prompt Design (2023)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2310.11324",{"title":580,"url":581},"\u002Fimages\u002Fblog\u002Fllm-model-migration-regression-testing\u002Fcover.webp","\u002Fimages\u002Fblog\u002Fllm-model-migration-regression-testing\u002Fog.jpg",[72,73,74],"LLM-Modellmigration ohne Regressionen: Deprecations, Evals, Shadow-Traffic","Ein Playbook für den Wechsel von LLM-Modellen: Deprecation-Fristen, Vertragsfixes, Regression-Suite, Prompt-Tuning, Shadow-Traffic und Canary.","LLM-Modellmigration: Umzugskisten auf einer Sackkarre fahren vom alten zum neuen Haus, dazwischen ein Maßband.",{"slug":661,"published":496,"minutes":662,"category":8,"tags":663,"tagPages":668,"keywords":670,"about":681,"sources":689,"cover":700,"og":701,"expertise":70,"locales":702,"lang":73,"title":703,"description":704,"coverAlt":705},"semantic-caching-llm-responses",6,[664,133,665,666,667],"Semantic cache","Latency","Embeddings","Redis",[669],"embeddings",[671,672,673,674,675,676,677,678,679,680],"semantisches caching llm","semantischer cache","llm-antwort-cache","gptcache","semantic cache ähnlichkeitsschwelle","prompt caching vs semantisches caching","llm-api-kosten senken","llm-antworten cachen redis","semantic cache falsch-positive","llm cache invalidierung",[682,685,688],{"name":683,"url":684},"Cache (computing)","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FCache_(computing)",{"name":686,"url":687},"Sentence embedding","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FSentence_embedding",{"name":341,"url":342},[690,693,696,697],{"title":691,"url":692},"Regmi and Pun: GPT Semantic Cache: Reducing LLM Costs and Latency via Semantic Embedding Caching (2024)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2411.05276",{"title":694,"url":695},"Zilliz: GPTCache README (modules, similarity evaluation, eviction)","https:\u002F\u002Fgithub.com\u002Fzilliztech\u002FGPTCache",{"title":305,"url":306},{"title":698,"url":699},"pgvector README: vector types and indexes","https:\u002F\u002Fgithub.com\u002Fpgvector\u002Fpgvector","\u002Fimages\u002Fblog\u002Fsemantic-caching-llm-responses\u002Fcover.webp","\u002Fimages\u002Fblog\u002Fsemantic-caching-llm-responses\u002Fog.jpg",[72,73,74],"Semantisches Caching für LLM-Apps: wann Antworten wiederverwenden sicher ist und wann nicht","Wie semantische Caches LLM-Antworten für ähnliche Fragen wiederverwenden, was sie vom Prompt-Caching trennt, wie man falsche Treffer und Lecks vermeidet.","Semantisches Caching: Eine Hand gibt an der Garderobe eine Marke ab und erhält einen Mantel.",{"slug":707,"published":708,"updated":709,"minutes":184,"category":8,"tags":710,"tagPages":715,"keywords":718,"about":727,"sources":737,"cover":807,"og":808,"expertise":70,"locales":809,"lang":73,"title":810,"description":811,"coverAlt":812},"self-hosted-llm-gdpr-cost","2026-10-09","2026-10-10",[711,712,713,714,130],"Self-hosting","GDPR","vLLM","GPU cost",[716,717],"self-hosting","gdpr",[719,720,721,722,723,724,725,726],"llm selbst hosten dsgvo","llm self-hosting datenschutz","gpu-speicher für llms","vllm durchsatz benchmark","eu gpu-cloud für llms","llm api vs self-hosting break-even","open-weight llm hardwarekosten","auftragsverarbeitungsvertrag llm",[728,731,732,735],{"name":729,"url":730},"General Data Protection Regulation","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FGeneral_Data_Protection_Regulation",{"name":341,"url":342},{"name":733,"url":734},"Graphics processing unit","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FGraphics_processing_unit",{"name":713,"url":736},"https:\u002F\u002Fgithub.com\u002Fvllm-project\u002Fvllm",[738,741,744,747,750,753,756,759,762,765,768,771,774,777,780,783,786,789,792,795,798,801,804],{"title":739,"url":740},"openai\u002Fgpt-oss-120b model card (Hugging Face)","https:\u002F\u002Fhuggingface.co\u002Fopenai\u002Fgpt-oss-120b",{"title":742,"url":743},"mistralai\u002FMistral-Small-3.2-24B-Instruct-2506 model card (Hugging Face)","https:\u002F\u002Fhuggingface.co\u002Fmistralai\u002FMistral-Small-3.2-24B-Instruct-2506",{"title":745,"url":746},"meta-llama\u002FLlama-3.3-70B-Instruct model card (Hugging Face)","https:\u002F\u002Fhuggingface.co\u002Fmeta-llama\u002FLlama-3.3-70B-Instruct",{"title":748,"url":749},"mistralai\u002FMistral-Large-3-675B-Instruct-2512 model card (Hugging Face)","https:\u002F\u002Fhuggingface.co\u002Fmistralai\u002FMistral-Large-3-675B-Instruct-2512",{"title":751,"url":752},"vLLM blog: v0.6.0 performance update (September 2024)","https:\u002F\u002Fblog.vllm.ai\u002F2024\u002F09\u002F05\u002Fperf-update.html",{"title":754,"url":755},"vLLM documentation: engine arguments","https:\u002F\u002Fdocs.vllm.ai\u002Fen\u002Flatest\u002Fconfiguration\u002Fengine_args.html",{"title":757,"url":758},"vLLM documentation: vllm bench serve","https:\u002F\u002Fdocs.vllm.ai\u002Fen\u002Flatest\u002Fcli\u002Fbench\u002Fserve.html",{"title":760,"url":761},"SemiAnalysis InferenceX: Llama 3.3 70B on H100 vs H200","https:\u002F\u002Finferencex.semianalysis.com\u002Fcompare\u002Fllama-3-3-70b-h100-vs-h200",{"title":763,"url":764},"Scaleway GPU instances pricing","https:\u002F\u002Fwww.scaleway.com\u002Fen\u002Fpricing\u002Fgpu\u002F",{"title":766,"url":767},"Scaleway blog: a transparent update on Scaleway pricing (27 April 2026)","https:\u002F\u002Fwww.scaleway.com\u002Fen\u002Fblog\u002Fa-transparent-update-on-scaleway-pricing\u002F",{"title":769,"url":770},"Scaleway Generative APIs pricing","https:\u002F\u002Fwww.scaleway.com\u002Fen\u002Fpricing\u002Fmodel-as-a-service\u002F",{"title":772,"url":773},"OVHcloud public cloud price list","https:\u002F\u002Fwww.ovhcloud.com\u002Fen\u002Fpublic-cloud\u002Fprices\u002F",{"title":775,"url":776},"OVHcloud AI Endpoints","https:\u002F\u002Fwww.ovhcloud.com\u002Fen\u002Fpublic-cloud\u002Fai-endpoints\u002F",{"title":778,"url":779},"OVHcloud AI Endpoints model catalogue","https:\u002F\u002Fwww.ovhcloud.com\u002Fen\u002Fpublic-cloud\u002Fai-endpoints\u002Fcatalog\u002F",{"title":781,"url":782},"Hetzner dedicated GPU servers","https:\u002F\u002Fwww.hetzner.com\u002Fdedicated-rootserver\u002Fmatrix-gpu\u002F",{"title":784,"url":785},"NVIDIA newsroom: GeForce RTX 50 Series launch","https:\u002F\u002Fnvidianews.nvidia.com\u002Fnews\u002Fnvidia-blackwell-geforce-rtx-50-series-opens-new-world-of-ai-computer-graphics",{"title":787,"url":788},"NVIDIA GeForce RTX 5090","https:\u002F\u002Fwww.nvidia.com\u002Fen-us\u002Fgeforce\u002Fgraphics-cards\u002F50-series\u002Frtx-5090\u002F",{"title":790,"url":791},"NVIDIA GeForce software licence","https:\u002F\u002Fwww.nvidia.com\u002Fen-us\u002Fdrivers\u002Fgeforce-license\u002F",{"title":793,"url":794},"GDPR Article 28: processor","https:\u002F\u002Fgdpr-info.eu\u002Fart-28-gdpr\u002F",{"title":796,"url":797},"GDPR Article 44: general principle for transfers","https:\u002F\u002Fgdpr-info.eu\u002Fart-44-gdpr\u002F",{"title":799,"url":800},"European Commission: EU-US data transfers (Data Privacy Framework)","https:\u002F\u002Fcommission.europa.eu\u002Flaw\u002Flaw-topic\u002Fdata-protection\u002Finternational-dimension-data-protection\u002Feu-us-data-transfers_en",{"title":802,"url":803},"EDPB Opinion 28\u002F2024 on AI models (adopted 17 December 2024)","https:\u002F\u002Fwww.edpb.europa.eu\u002Fsystem\u002Ffiles\u002F2024-12\u002Fedpb_opinion_202428_ai-models_en.pdf",{"title":805,"url":806},"Mistral AI pricing","https:\u002F\u002Fmistral.ai\u002Fpricing","\u002Fimages\u002Fblog\u002Fself-hosted-llm-gdpr-cost\u002Fcover.webp","\u002Fimages\u002Fblog\u002Fself-hosted-llm-gdpr-cost\u002Fog.jpg",[72,73,74],"LLMs selbst hosten für die DSGVO: wann es nötig ist und was es kostet","LLMs selbst hosten für die DSGVO: wann es nötig ist, GPU-Speicher für offene Modelle, EU-Preise Stand Oktober 2026 und Break-even pro Million Token.","Self-Hosting von LLMs: Ein Serverschrank in einem verschlossenen Haus neben Sparschwein und Münzen.",{"slug":814,"published":815,"minutes":253,"category":8,"tags":816,"tagPages":822,"keywords":824,"about":832,"sources":843,"cover":897,"og":898,"expertise":70,"locales":899,"lang":73,"title":900,"description":901,"coverAlt":902},"deepeval","2026-10-06",[817,818,819,820,821],"LLM evaluation","pytest","LLM-as-a-judge","Red teaming","Open source",[823],"open-source",[814,825,826,827,828,829,830,831],"deepeval vs ragas vergleich","llm evaluation framework","pytest für llm-ausgaben","g-eval metrik","llm-judge kosten","deepeval kosten","open-source llm red teaming",[833,836,839,842],{"name":834,"url":835},"DeepEval","https:\u002F\u002Fdeepeval.com",{"name":837,"url":838},"Confident AI","https:\u002F\u002Fwww.confident-ai.com",{"name":840,"url":841},"Pytest","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FPytest",{"name":341,"url":342},[844,847,850,853,856,859,862,864,867,870,873,876,879,882,885,888,891,894],{"title":845,"url":846},"DeepEval documentation: getting started","https:\u002F\u002Fdeepeval.com\u002Fdocs\u002Fgetting-started",{"title":848,"url":849},"GitHub: confident-ai\u002Fdeepeval, the README and licence","https:\u002F\u002Fgithub.com\u002Fconfident-ai\u002Fdeepeval",{"title":851,"url":852},"PyPI: deepeval, the latest release and Python requirement","https:\u002F\u002Fpypi.org\u002Fproject\u002Fdeepeval\u002F",{"title":854,"url":855},"DeepEval documentation: metrics introduction","https:\u002F\u002Fdeepeval.com\u002Fdocs\u002Fmetrics-introduction",{"title":857,"url":858},"DeepEval documentation: G-Eval","https:\u002F\u002Fdeepeval.com\u002Fdocs\u002Fmetrics-llm-evals",{"title":860,"url":861},"DeepEval documentation: Faithfulness","https:\u002F\u002Fdeepeval.com\u002Fdocs\u002Fmetrics-faithfulness",{"title":863,"url":593},"DeepEval documentation: Tool Correctness",{"title":865,"url":866},"DeepEval documentation: generate goldens from documents","https:\u002F\u002Fdeepeval.com\u002Fdocs\u002Fsynthesizer-generate-from-docs",{"title":868,"url":869},"DeepEval documentation: unit testing in CI\u002FCD","https:\u002F\u002Fdeepeval.com\u002Fdocs\u002Fevaluation-unit-testing-in-ci-cd",{"title":871,"url":872},"DeepEval FAQ","https:\u002F\u002Fdeepeval.com\u002Fdocs\u002Ffaq",{"title":874,"url":875},"DeepEval documentation: data privacy","https:\u002F\u002Fdeepeval.com\u002Fdocs\u002Fdata-privacy",{"title":877,"url":878},"GitHub: confident-ai\u002Fdeepteam, the red-teaming framework","https:\u002F\u002Fgithub.com\u002Fconfident-ai\u002Fdeepteam",{"title":880,"url":881},"Confident AI pricing","https:\u002F\u002Fwww.confident-ai.com\u002Fpricing",{"title":883,"url":884},"Confident AI documentation: data residency","https:\u002F\u002Fwww.confident-ai.com\u002Fdocs\u002Fsettings\u002Fdata-residency",{"title":886,"url":887},"Confident AI subprocessor list","https:\u002F\u002Fwww.confident-ai.com\u002Fsubprocessors-list",{"title":889,"url":890},"GitHub: explodinggradients\u002Fragas, the README","https:\u002F\u002Fgithub.com\u002Fexplodinggradients\u002Fragas",{"title":892,"url":893},"GitHub: promptfoo\u002Fpromptfoo, the README","https:\u002F\u002Fgithub.com\u002Fpromptfoo\u002Fpromptfoo",{"title":895,"url":896},"Braintrust pricing","https:\u002F\u002Fwww.braintrust.dev\u002Fpricing","\u002Fimages\u002Fblog\u002Fdeepeval\u002Fcover.webp","\u002Fimages\u002Fblog\u002Fdeepeval\u002Fog.jpg",[72,73,74],"DeepEval-Test: pytest für LLM-Ausgaben, samt Richter-Rechnung","DeepEval führt LLM-Prüfungen als pytest-artige Tests aus, mit eingebauten Richter-Metriken. Die Bibliothek ist frei; Richter-Aufrufe und Datenfluss kosten.","DeepEval: Ein Richterhammer neben einem Reagenzglasständer und einer offenen Geldbörse.",{"slug":904,"published":815,"minutes":253,"category":8,"tags":905,"tagPages":911,"keywords":913,"about":921,"sources":932,"cover":1008,"og":1009,"expertise":70,"locales":1010,"lang":73,"title":1011,"description":1012,"coverAlt":1013},"dspy",[906,907,908,909,910],"Prompt optimisation","LLM programs","MIPROv2","GEPA","Python",[912],"python",[904,914,915,916,917,918,919,920],"dspy anleitung","dspy optimierer","miprov2","gepa prompt-optimierung","dspy vs prompt engineering vergleich","automatische prompt-optimierung","dspy signaturen",[922,925,926,929],{"name":923,"url":924},"DSPy","https:\u002F\u002Fdspy.ai",{"name":274,"url":275},{"name":927,"url":928},"Bayesian optimization","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FBayesian_optimization",{"name":930,"url":931},"Fine-tuning (deep learning)","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FFine-tuning_(deep_learning)",[933,936,939,942,945,948,951,954,957,960,963,966,969,972,975,978,981,984,987,990,993,996,999,1002,1005],{"title":934,"url":935},"DSPy home and cost example","https:\u002F\u002Fdspy.ai\u002Fcurrent\u002F",{"title":937,"url":938},"DSPy installation","https:\u002F\u002Fdspy.ai\u002Fcurrent\u002Fgetting-started\u002Finstallation\u002F",{"title":940,"url":941},"DSPy: program, don't prompt","https:\u002F\u002Fdspy.ai\u002Fcurrent\u002Fgetting-started\u002Fprogram-dont-prompt\u002F",{"title":943,"url":944},"DSPy signatures","https:\u002F\u002Fdspy.ai\u002Fcurrent\u002Fdiving-deeper\u002Fsignatures-in-depth\u002F",{"title":946,"url":947},"DSPy metrics","https:\u002F\u002Fdspy.ai\u002Fcurrent\u002Fgetting-started\u002Fmetrics\u002F",{"title":949,"url":950},"DSPy Example API","https:\u002F\u002Fdspy.ai\u002Fcurrent\u002Fapi\u002Fprimitives\u002FExample\u002F",{"title":952,"url":953},"DSPy Predict API","https:\u002F\u002Fdspy.ai\u002Fcurrent\u002Fapi\u002Fmodules\u002FPredict\u002F",{"title":955,"url":956},"DSPy ChainOfThought API","https:\u002F\u002Fdspy.ai\u002Fcurrent\u002Fapi\u002Fmodules\u002FChainOfThought\u002F",{"title":958,"url":959},"DSPy ReAct API","https:\u002F\u002Fdspy.ai\u002Fcurrent\u002Fapi\u002Fmodules\u002FReAct\u002F",{"title":961,"url":962},"DSPy optimizers index","https:\u002F\u002Fdspy.ai\u002Fcurrent\u002Fapi\u002Foptimizers\u002F",{"title":964,"url":965},"DSPy MIPROv2 API","https:\u002F\u002Fdspy.ai\u002Fcurrent\u002Fapi\u002Foptimizers\u002FMIPROv2\u002F",{"title":967,"url":968},"MIPROv2 source code","https:\u002F\u002Fgithub.com\u002Fstanfordnlp\u002Fdspy\u002Fblob\u002Fmain\u002Fdspy\u002Fteleprompt\u002Fmipro_optimizer_v2.py",{"title":970,"url":971},"DSPy BootstrapFewShot API","https:\u002F\u002Fdspy.ai\u002Fcurrent\u002Fapi\u002Foptimizers\u002FBootstrapFewShot\u002F",{"title":973,"url":974},"DSPy BootstrapFinetune API","https:\u002F\u002Fdspy.ai\u002Fcurrent\u002Fapi\u002Foptimizers\u002FBootstrapFinetune\u002F",{"title":976,"url":977},"DSPy GEPA guide","https:\u002F\u002Fdspy.ai\u002Fcurrent\u002Fgetting-started\u002Fgepa-optimization\u002F",{"title":979,"url":980},"GEPA source code","https:\u002F\u002Fgithub.com\u002Fstanfordnlp\u002Fdspy\u002Fblob\u002Fmain\u002Fdspy\u002Fteleprompt\u002Fgepa\u002Fgepa.py",{"title":982,"url":983},"DSPy saving programs","https:\u002F\u002Fdspy.ai\u002Fcurrent\u002Ftutorials\u002Fsaving\u002F",{"title":985,"url":986},"DSPy caching","https:\u002F\u002Fdspy.ai\u002Fcurrent\u002Ftutorials\u002Fcache\u002F",{"title":988,"url":989},"DSPy LM API","https:\u002F\u002Fdspy.ai\u002Fcurrent\u002Fapi\u002Fmodels\u002FLM\u002F",{"title":991,"url":992},"DSPy FAQ","https:\u002F\u002Fdspy.ai\u002Fcurrent\u002Ffaqs\u002F",{"title":994,"url":995},"DSPy GitHub repository","https:\u002F\u002Fgithub.com\u002Fstanfordnlp\u002Fdspy",{"title":997,"url":998},"DSPy 3.4.0 release","https:\u002F\u002Fgithub.com\u002Fstanfordnlp\u002Fdspy\u002Freleases\u002Ftag\u002F3.4.0",{"title":1000,"url":1001},"DSPy paper, arXiv","https:\u002F\u002Farxiv.org\u002Fabs\u002F2310.03714",{"title":1003,"url":1004},"MIPRO paper, arXiv","https:\u002F\u002Farxiv.org\u002Fabs\u002F2406.11695",{"title":1006,"url":1007},"GEPA paper, arXiv","https:\u002F\u002Farxiv.org\u002Fabs\u002F2507.19457","\u002Fimages\u002Fblog\u002Fdspy\u002Fcover.webp","\u002Fimages\u002Fblog\u002Fdspy\u002Fog.jpg",[72,73,74],"DSPy im Test: Prompts gegen eine Metrik kompilieren, nicht von Hand","DSPy erzeugt Prompts aus Signaturen, einer Metrik und Beispielen. Was Optimierer an Modellaufrufen kosten und wann ein handgeschriebener Prompt besser ist.","DSPy: Ein Roboter kurbelt eine Maschine, die ein zerknülltes Blatt glättet, darüber eine Zielscheibe.",{"slug":1015,"published":1016,"updated":709,"minutes":253,"category":8,"tags":1017,"tagPages":1023,"keywords":1024,"about":1032,"sources":1040,"cover":1078,"og":1079,"expertise":70,"locales":1080,"lang":73,"title":1081,"description":1082,"coverAlt":1083},"llama-cpp","2026-10-05",[1018,1019,1020,1021,1022],"llama.cpp","GGUF","Quantisation","Local inference","llama-server",[],[1018,1025,1026,1027,1028,1029,1030,1031],"llama.cpp test","gguf quantisierung stufen","llama-server openai-kompatibel","llama.cpp vs ollama vergleich","llm lokal betreiben dsgvo","llama.cpp gpu-backends","q4_k_m vs q8_0 qualität",[1033,1035,1036,1039],{"name":1018,"url":1034},"https:\u002F\u002Fgithub.com\u002Fggml-org\u002Fllama.cpp",{"name":341,"url":342},{"name":1037,"url":1038},"Quantization (signal processing)","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FQuantization_(signal_processing)",{"name":729,"url":730},[1041,1043,1046,1049,1052,1054,1057,1060,1063,1066,1069,1071,1074,1075],{"title":1042,"url":1034},"llama.cpp repository: goals, backends, licence",{"title":1044,"url":1045},"llama.cpp releases: builds b11538 to b11541","https:\u002F\u002Fgithub.com\u002Fggml-org\u002Fllama.cpp\u002Freleases",{"title":1047,"url":1048},"llama.cpp build documentation: CMake flags","https:\u002F\u002Fgithub.com\u002Fggml-org\u002Fllama.cpp\u002Fblob\u002Fmaster\u002Fdocs\u002Fbuild.md",{"title":1050,"url":1051},"llama.cpp server README: endpoints, slots and grammars","https:\u002F\u002Fgithub.com\u002Fggml-org\u002Fllama.cpp\u002Fblob\u002Fmaster\u002Ftools\u002Fserver\u002FREADME.md",{"title":1053,"url":430},"llama.cpp quantisation README: bits, size and speed",{"title":1055,"url":1056},"GGUF specification in the ggml repository","https:\u002F\u002Fgithub.com\u002Fggml-org\u002Fggml\u002Fblob\u002Fmaster\u002Fdocs\u002Fgguf.md",{"title":1058,"url":1059},"Performance of llama.cpp on Apple Silicon M-series","https:\u002F\u002Fgithub.com\u002Fggml-org\u002Fllama.cpp\u002Fdiscussions\u002F4167",{"title":1061,"url":1062},"Performance of llama.cpp on Nvidia CUDA","https:\u002F\u002Fgithub.com\u002Fggml-org\u002Fllama.cpp\u002Fdiscussions\u002F15013",{"title":1064,"url":1065},"Ollama README: supported backends and REST API","https:\u002F\u002Fgithub.com\u002Follama\u002Follama",{"title":1067,"url":1068},"LM Studio documentation: app overview","https:\u002F\u002Flmstudio.ai\u002Fdocs\u002Fapp",{"title":1070,"url":736},"vLLM README: features, hardware and licence",{"title":1072,"url":1073},"vLLM documentation: GGUF support","https:\u002F\u002Fdocs.vllm.ai\u002Fen\u002Flatest\u002Ffeatures\u002Fquantization\u002Fgguf.html",{"title":793,"url":794},{"title":1076,"url":1077},"GDPR Article 32: security of processing","https:\u002F\u002Fgdpr-info.eu\u002Fart-32-gdpr\u002F","\u002Fimages\u002Fblog\u002Fllama-cpp\u002Fcover.webp","\u002Fimages\u002Fblog\u002Fllama-cpp\u002Fog.jpg",[72,73,74],"llama.cpp im Test: die lokale Engine unter Ollama und LM Studio","llama.cpp führt offene Modelle in reinem C und C++ auf Metal, CUDA, Vulkan oder der CPU aus. Ich zeige GGUF-Quants, llama-server und die Grenzen.","llama.cpp: Ein Lama mit Zahnrad steht auf einem Schreibtisch neben einem PC.",{"slug":1085,"published":1016,"minutes":1086,"category":8,"tags":1087,"tagPages":1092,"keywords":1097,"about":1105,"sources":1117,"cover":1187,"og":1188,"expertise":70,"locales":1189,"lang":73,"title":1190,"description":1191,"coverAlt":1192},"opik",7,[1088,1089,1090,1091,711],"LLM observability","Tracing","Evaluation","OpenTelemetry",[1093,1094,1095,1096,716],"llm-observability","tracing","evaluation","opentelemetry",[1085,1098,1099,1100,1101,1102,1103,1104],"opik test","opik vs langfuse vergleich","opik kosten","self-hosted llm-tracing","opik opentelemetry anbindung","llm-as-a-judge metriken","opik guardrails einrichten",[1106,1109,1111,1114],{"name":1107,"url":1108},"Comet ML","https:\u002F\u002Fwww.comet.com",{"name":1091,"url":1110},"https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FOpenTelemetry",{"name":1112,"url":1113},"Observability (software)","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FObservability_(software)",{"name":1115,"url":1116},"Apache License","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FApache_License",[1118,1121,1124,1127,1130,1133,1136,1139,1142,1145,1148,1151,1154,1157,1160,1163,1166,1169,1172,1175,1178,1181,1184],{"title":1119,"url":1120},"Opik repository README on GitHub","https:\u002F\u002Fgithub.com\u002Fcomet-ml\u002Fopik",{"title":1122,"url":1123},"Opik LICENSE file: Apache License 2.0","https:\u002F\u002Fgithub.com\u002Fcomet-ml\u002Fopik\u002Fblob\u002Fmain\u002FLICENSE",{"title":1125,"url":1126},"Opik on PyPI: package metadata","https:\u002F\u002Fpypi.org\u002Fpypi\u002Fopik\u002Fjson",{"title":1128,"url":1129},"Opik product page","https:\u002F\u002Fwww.comet.com\u002Fsite\u002Fproducts\u002Fopik\u002F",{"title":1131,"url":1132},"Opik pricing","https:\u002F\u002Fwww.comet.com\u002Fsite\u002Fpricing\u002F",{"title":1134,"url":1135},"Opik docs: run locally with Docker Compose","https:\u002F\u002Fwww.comet.com\u002Fdocs\u002Fopik\u002Fself-host\u002Flocal_deployment",{"title":1137,"url":1138},"Opik docs: Kubernetes deployment with Helm","https:\u002F\u002Fwww.comet.com\u002Fdocs\u002Fopik\u002Fself-host\u002Fkubernetes",{"title":1140,"url":1141},"Opik docs: platform architecture","https:\u002F\u002Fwww.comet.com\u002Fdocs\u002Fopik\u002Fself-host\u002Farchitecture",{"title":1143,"url":1144},"Opik docs: tracing getting started","https:\u002F\u002Fwww.comet.com\u002Fdocs\u002Fopik\u002Ftracing\u002Fgetting-started",{"title":1146,"url":1147},"Opik docs: OpenTelemetry integration","https:\u002F\u002Fwww.comet.com\u002Fdocs\u002Fopik\u002Fintegrations\u002Fopentelemetry",{"title":1149,"url":1150},"Opik docs: evaluation metrics overview","https:\u002F\u002Fwww.comet.com\u002Fdocs\u002Fopik\u002Fevaluation\u002Fmetrics\u002Foverview",{"title":1152,"url":1153},"Opik docs: online evaluation rules","https:\u002F\u002Fwww.comet.com\u002Fdocs\u002Fopik\u002Fproduction\u002Fonline-evaluation\u002Frules",{"title":1155,"url":1156},"Opik docs: Prompt Library overview","https:\u002F\u002Fwww.comet.com\u002Fdocs\u002Fopik\u002Fdevelopment\u002Fprompt-library\u002Foverview",{"title":1158,"url":1159},"Opik docs: optimisation algorithms","https:\u002F\u002Fwww.comet.com\u002Fdocs\u002Fopik\u002Fdevelopment\u002Foptimization-runs\u002Falgorithms\u002Foverview",{"title":1161,"url":1162},"Opik docs: guardrails overview","https:\u002F\u002Fwww.comet.com\u002Fdocs\u002Fopik\u002Fguardrails\u002Foverview",{"title":1164,"url":1165},"Opik docs: guardrails server","https:\u002F\u002Fwww.comet.com\u002Fdocs\u002Fopik\u002Fguardrails\u002Fserver",{"title":1167,"url":1168},"Opik docs: SDK anonymizers","https:\u002F\u002Fwww.comet.com\u002Fdocs\u002Fopik\u002Fproduction\u002Fgateway-guardrails\u002Fanonymizers",{"title":1170,"url":1171},"Opik docs: data anonymization","https:\u002F\u002Fwww.comet.com\u002Fdocs\u002Fopik\u002Fadministration\u002Fdata_anonymization",{"title":1173,"url":1174},"Comet ML privacy policy","https:\u002F\u002Fwww.comet.com\u002Fsite\u002Fprivacy-policy\u002F",{"title":1176,"url":1177},"Comet Trust Center","https:\u002F\u002Ftrust.comet.com\u002F",{"title":1179,"url":1180},"Langfuse repository README (licence)","https:\u002F\u002Fgithub.com\u002Flangfuse\u002Flangfuse",{"title":1182,"url":1183},"Arize Phoenix repository README (licence)","https:\u002F\u002Fgithub.com\u002FArize-ai\u002Fphoenix",{"title":1185,"url":1186},"LangChain pricing (LangSmith plans)","https:\u002F\u002Fwww.langchain.com\u002Fpricing","\u002Fimages\u002Fblog\u002Fopik\u002Fcover.webp","\u002Fimages\u002Fblog\u002Fopik\u002Fog.jpg",[72,73,74],"Opik im Test: Open-Source-Tracing und Evals mit US-gehosteter Cloud","Opik im Test: Traces, Judge-Metriken, Prompts und Optimierer unter Apache 2.0. Selbst hosten ist gratis, Pro-Cloud ab 19 Dollar, Hosting in den USA.","Opik: Ein Fernglas auf einem Stapel Notizbücher, im Hintergrund eine Wolke über einer Fahne.",{"slug":1194,"published":1195,"minutes":7,"category":8,"tags":1196,"tagPages":1202,"keywords":1203,"about":1214,"sources":1224,"cover":1246,"og":1247,"expertise":70,"locales":1248,"lang":73,"title":1249,"description":1250,"coverAlt":1251},"local-text-to-speech-pipeline","2026-10-01",[1197,1198,1199,1200,1201],"Text-to-speech","Audio","Kokoro","FFmpeg","Vue",[],[1204,1205,1206,1207,1208,1209,1210,1211,1212,1213],"lokales text-to-speech","tts-pipeline lokal","kokoro tts","piper tts","bark tts vergleich","blogartikel vertonen","wav zu m4a ffmpeg","aac 192 kbit\u002Fs faststart einstellen","vue audio-player","onnx tts mac einrichten",[1215,1218,1221],{"name":1216,"url":1217},"Speech synthesis","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FSpeech_synthesis",{"name":1219,"url":1220},"Advanced Audio Coding","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FAdvanced_Audio_Coding",{"name":1222,"url":1223},"ESpeak","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FESpeak",[1225,1228,1231,1234,1237,1240,1243],{"title":1226,"url":1227},"Kokoro onnx: runtime bindings","https:\u002F\u002Fgithub.com\u002Fthewh1teagle\u002Fkokoro-onnx",{"title":1229,"url":1230},"Kokoro-82M model card","https:\u002F\u002Fhuggingface.co\u002Fhexgrad\u002FKokoro-82M",{"title":1232,"url":1233},"Piper text-to-speech","https:\u002F\u002Fgithub.com\u002Frhasspy\u002Fpiper",{"title":1235,"url":1236},"Bark by Suno","https:\u002F\u002Fgithub.com\u002Fsuno-ai\u002Fbark",{"title":1238,"url":1239},"FFmpeg AAC encoder documentation","https:\u002F\u002Fffmpeg.org\u002Fffmpeg-codecs.html#aac",{"title":1241,"url":1242},"torch.load weights_only documentation","https:\u002F\u002Fpytorch.org\u002Fdocs\u002Fstable\u002Fgenerated\u002Ftorch.load.html",{"title":1244,"url":1245},"ESpeak NG phonemizer","https:\u002F\u002Fgithub.com\u002Fespeak-ng\u002Fespeak-ng","\u002Fimages\u002Fblog\u002Flocal-text-to-speech-pipeline\u002Fcover.webp","\u002Fimages\u002Fblog\u002Flocal-text-to-speech-pipeline\u002Fog.jpg",[72,73,74],"Lokales Text-to-Speech im großen Maßstab: 96 Artikel mit offenen Modellen vertont","Drei offene TTS-Modelle, ein Laptop: 96 Artikel zu 1.512 Minuten Sprache, chunkierte Synthese, 192-kbit\u002Fs-AAC und ein Player, der mitläuft.","Lokales Text-to-Speech: Ein Bücherstapel neben einem kleinen Rechner mit einer Reihe Kopfhörer an Haken.",{"slug":1253,"published":1254,"minutes":7,"category":8,"tags":1255,"tagPages":1258,"keywords":1259,"about":1267,"sources":1271,"cover":1289,"og":1290,"expertise":70,"locales":1291,"lang":73,"title":1292,"description":1293,"coverAlt":1294},"ollama","2026-09-29",[1021,1256,1018,1019,1257],"Open models","Model serving",[],[1253,1260,1261,1262,1263,1264,1265,1266],"ollama test","ollama vs lm studio vergleich","ollama vs vllm vergleich","lokale llm laufzeit","gguf modellserver","ollama selbst hosten","ollama api anleitung",[1268],{"name":1269,"url":1270},"Ollama (software)","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FOllama",[1272,1275,1277,1280,1283,1286],{"title":1273,"url":1274},"Ollama API documentation","https:\u002F\u002Fdocs.ollama.com\u002Fapi",{"title":1276,"url":1065},"Ollama on GitHub, with the MIT LICENSE file",{"title":1278,"url":1279},"Ollama terms of service, last updated May 2026","https:\u002F\u002Follama.com\u002Fterms",{"title":1281,"url":1282},"Ollama pricing, cloud plans and per-token model rates","https:\u002F\u002Follama.com\u002Fpricing",{"title":1284,"url":1285},"Hardware support: Nvidia, AMD, Metal and Vulkan","https:\u002F\u002Fdocs.ollama.com\u002Fgpu",{"title":1287,"url":1288},"OpenAI compatibility, including what is not supported","https:\u002F\u002Fdocs.ollama.com\u002Fapi\u002Fopenai-compatibility","\u002Fimages\u002Fblog\u002Follama\u002Fcover.webp","\u002Fimages\u002Fblog\u002Follama\u002Fog.jpg",[72,73,74],"Ollama im Test: der freundliche Weg zu offenen Modellen","Ollama im Test: offene Modelle per HTTP-API auf eigener Hardware. Was es gut macht, wo der Durchsatz zu kurz kommt und was die MIT-Lizenz nicht abdeckt.","Ollama: Ein Lama sitzt mit dem Laptop gemütlich auf einem Sofa.",{"slug":1296,"published":1297,"minutes":7,"category":8,"tags":1298,"tagPages":1303,"keywords":1305,"about":1313,"sources":1320,"cover":1372,"og":1373,"expertise":70,"locales":1374,"lang":73,"title":1375,"description":1376,"coverAlt":1377},"portkey","2026-09-28",[190,1299,1300,1301,1302],"Guardrails","Routing","Observability","Cost control",[1304],"guardrails",[1296,1306,1307,1308,1309,1310,1311,1312],"portkey ki-gateway","portkey vs litellm vergleich","llm-gateway vergleich","llm-gateway latenz-overhead","llm-guardrails im gateway","portkey selbst hosten","portkey preise",[1314,1317],{"name":1315,"url":1316},"Portkey","https:\u002F\u002Fportkey.ai",{"name":1318,"url":1319},"API gateway","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FAPI_gateway",[1321,1324,1327,1330,1333,1336,1339,1342,1345,1348,1351,1354,1357,1360,1363,1366,1369],{"title":1322,"url":1323},"Portkey docs: AI Gateway","https:\u002F\u002Fportkey.ai\u002Fdocs\u002Fproduct\u002Fai-gateway",{"title":1325,"url":1326},"Portkey docs: Getting started with the AI Gateway","https:\u002F\u002Fdocs.portkey.ai\u002Fdocs\u002Fguides\u002Fgetting-started\u002Fgetting-started-with-ai-gateway",{"title":1328,"url":1329},"Portkey docs: Gateway config object","https:\u002F\u002Fportkey.ai\u002Fdocs\u002Fapi-reference\u002Fconfig-object",{"title":1331,"url":1332},"Portkey docs: Guardrails","https:\u002F\u002Fportkey.ai\u002Fdocs\u002Fproduct\u002Fguardrails",{"title":1334,"url":1335},"Portkey docs: Guardrail endpoints and capabilities","https:\u002F\u002Fportkey.ai\u002Fdocs\u002Fproduct\u002Fguardrails\u002Fcapabilities",{"title":1337,"url":1338},"Portkey docs: Cache, simple and semantic","https:\u002F\u002Fportkey.ai\u002Fdocs\u002Fproduct\u002Fai-gateway\u002Fcache-simple-and-semantic",{"title":1340,"url":1341},"Portkey docs: Load balancing","https:\u002F\u002Fportkey.ai\u002Fdocs\u002Fproduct\u002Fai-gateway\u002Fload-balancing",{"title":1343,"url":1344},"Portkey docs: Enterprise hybrid deployment architecture","https:\u002F\u002Fportkey.ai\u002Fdocs\u002Fself-hosting\u002Fhybrid-deployments\u002Farchitecture",{"title":1346,"url":1347},"Portkey pricing","https:\u002F\u002Fportkey.ai\u002Fpricing",{"title":1349,"url":1350},"Portkey gateway on GitHub, MIT licensed","https:\u002F\u002Fgithub.com\u002FPortkey-AI\u002Fgateway",{"title":1352,"url":1353},"Portkey's own benchmark: gateway versus direct Bedrock","https:\u002F\u002Fgithub.com\u002FPortkey-AI\u002Fbenchmark-test",{"title":1355,"url":1356},"Portkey status page","https:\u002F\u002Fstatus.portkey.ai\u002F",{"title":1358,"url":1359},"Palo Alto Networks completes acquisition of Portkey, May 2026","https:\u002F\u002Fwww.paloaltonetworks.com\u002Fcompany\u002Fpress\u002F2026\u002Fpalo-alto-networks-completes-acquisition-of-portkey-to-secure-ai-agents",{"title":1361,"url":1362},"Palo Alto Networks: Prisma AIRS AI Gateway","https:\u002F\u002Fwww.paloaltonetworks.com\u002Fai-security\u002Fai-gateway",{"title":1364,"url":1365},"Cloudflare AI Gateway pricing","https:\u002F\u002Fdevelopers.cloudflare.com\u002Fai-gateway\u002Freference\u002Fpricing\u002F",{"title":1367,"url":1368},"LiteLLM pricing","https:\u002F\u002Fwww.litellm.ai\u002Fpricing",{"title":1370,"url":1371},"OpenRouter pricing","https:\u002F\u002Fopenrouter.ai\u002Fpricing","\u002Fimages\u002Fblog\u002Fportkey\u002Fcover.webp","\u002Fimages\u002Fblog\u002Fportkey\u002Fog.jpg",[72,73,74],"Portkey: ein LLM-Gateway für die Produktion, geprüft auf Routing, Guardrails und Kosten","Portkey im Test: Retries, Fallbacks, Cache, Guardrails und Kostenkontrolle hinter einer OpenAI-kompatiblen Schnittstelle. Was das Gateway an Latenz kostet.","Portkey: Ein Hafentor mit Leuchtturm lenkt kleine Boote in verschiedene Kanäle.",{"slug":1379,"published":1380,"updated":1297,"minutes":253,"category":8,"tags":1381,"tagPages":1385,"keywords":1386,"about":1398,"sources":1406,"cover":1424,"og":1425,"expertise":70,"locales":1426,"lang":73,"title":1427,"description":1428,"coverAlt":1429},"artificial-analysis-leaderboard-claude-opus-5-5","2026-09-23",[1382,1383,1384,133],"Claude Opus 5.5","Artificial Analysis","LLM benchmarks",[],[1387,1388,1389,1390,1391,1392,1393,1394,1395,1396,1397],"claude opus 5.5 test","claude opus 5.5 preis","claude opus 5.5","artificial analysis intelligence index","ki-modell rangliste","opus 5.5 vs gpt-6 astra","llm kosten pro aufgabe","reasoning-effort einstellung","opus 5.5 kosten vergleich","bestes llm september 2026","claude opus 5.5 effort medium",[1399,1402,1403],{"name":1400,"url":1401},"Claude (language model)","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FClaude_(language_model)",{"name":341,"url":342},{"name":1404,"url":1405},"Benchmark (computing)","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FBenchmark_(computing)",[1407,1410,1413,1416,1419,1422],{"title":1408,"url":1409},"Artificial Analysis: Claude Opus 5.5 takes the top spot on the Artificial Analysis Intelligence Index (22 September 2026)","https:\u002F\u002Fartificialanalysis.ai\u002Farticles\u002Fclaude-opus-5-5",{"title":1411,"url":1412},"Artificial Analysis: Claude Opus 5.5 (max) model page","https:\u002F\u002Fartificialanalysis.ai\u002Fmodels\u002Fclaude-opus-5-5",{"title":1414,"url":1415},"Artificial Analysis: Claude Opus 5.5 (medium) model page","https:\u002F\u002Fartificialanalysis.ai\u002Fmodels\u002Fclaude-opus-5-5-medium",{"title":1417,"url":1418},"Artificial Analysis: Claude Opus 5 (max) model page","https:\u002F\u002Fartificialanalysis.ai\u002Fmodels\u002Fclaude-opus-5",{"title":1420,"url":1421},"OfficeChai: Claude Opus 5.5 creates a 5-point lead over GPT-6 Astra","https:\u002F\u002Fofficechai.com\u002Fai\u002Fclaude-opus-5-5-creates-5-point-lead-over-gpt-6-astra-jumps-to-top-spot-on-artificial-analysis-intelligence-index\u002F",{"title":1423,"url":118},"Claude API docs: Models overview, context windows and prices (as of September 2026)","\u002Fimages\u002Fblog\u002Fartificial-analysis-leaderboard-claude-opus-5-5\u002Fcover.webp","\u002Fimages\u002Fblog\u002Fartificial-analysis-leaderboard-claude-opus-5-5\u002Fog.jpg",[72,73,74],"Claude Opus 5.5 holt Platz 1 bei Artificial Analysis – die eigentliche Story ist medium","Claude Opus 5.5 ist Platz 1 von 211 Modellen bei Artificial Analysis (58 Punkte). Mit Effort medium erreicht es Opus 5 für 1,34 $ statt 5,86 $ pro Aufgabe.","Claude Opus 5.5: Ein Pokal auf der obersten Stufe eines Podests, daneben ein Regler in Mittelstellung.",{"slug":1431,"published":1432,"minutes":184,"category":8,"tags":1433,"tagPages":1439,"keywords":1440,"about":1450,"sources":1459,"cover":1477,"og":1478,"expertise":70,"locales":1479,"lang":73,"title":1480,"description":1481,"coverAlt":1482},"jev-typed-decisions-llm-routing","2026-09-17",[1434,1435,1436,1437,1438],"LLM routing","Classification","Calibration","OpenRouter","Human in the loop",[],[1441,1442,1443,1444,1445,1446,1447,1448,1449],"routing für llm-anfragen","llm-klassifikation","kalibrierte konfidenz llm","typisierte llm-ausgabe","jev modell","openrouter decisions api nutzen","ki-triage code-review-befunde","llm-antworten mit niedriger konfidenz an mensch","llm-klassifikator vs structured output",[1451,1454,1457],{"name":1452,"url":1453},"Calibration (statistics)","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FCalibration_(statistics)",{"name":1455,"url":1456},"Statistical classification","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FStatistical_classification",{"name":1437,"url":1458},"https:\u002F\u002Fopenrouter.ai\u002F",[1460,1463,1466,1469,1472,1474],{"title":1461,"url":1462},"OpenRouter: Jev documentation","https:\u002F\u002Fopenrouter.ai\u002Fdocs\u002Fguides\u002Fcommunity\u002Fjev",{"title":1464,"url":1465},"OpenRouter: What is Jev?","https:\u002F\u002Fopenrouter.ai\u002Fblog\u002Finsights\u002Fwhat-is-jev\u002F",{"title":1467,"url":1468},"OpenRouter: How to use Jev","https:\u002F\u002Fopenrouter.ai\u002Fblog\u002Ftutorials\u002Fhow-to-use-jev\u002F",{"title":1470,"url":1471},"Guo et al.: On Calibration of Modern Neural Networks","https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.04599",{"title":1473,"url":529},"Claude API: Structured outputs",{"title":1475,"url":1476},"AI SDK: Generating structured data","https:\u002F\u002Fai-sdk.dev\u002Fdocs\u002Fai-sdk-core\u002Fgenerating-structured-data","\u002Fimages\u002Fblog\u002Fjev-typed-decisions-llm-routing\u002Fcover.webp","\u002Fimages\u002Fblog\u002Fjev-typed-decisions-llm-routing\u002Fog.jpg",[72,73,74],"Typisierte Entscheidungen für LLM-Routing und Triage: kalibrierte Konfidenz mit Jev","LLM-Routing mit typisierten Entscheidungen in Jev: Choice-, Score- und Ja-Wahrscheinlichkeits-Antworten mit kalibrierter Konfidenz, Schwellen und Hand-off.","LLM-Routing mit Jev: Eine Hand sortiert einen Umschlag in eines von drei Fächern, daneben eine Anzeige.",{"slug":1484,"published":1485,"updated":1486,"minutes":184,"category":8,"tags":1487,"tagPages":1488,"keywords":1489,"about":1497,"sources":1505,"cover":1529,"og":1530,"expertise":70,"locales":1531,"lang":73,"title":1532,"description":1533,"coverAlt":1534},"langfuse","2026-08-13","2026-10-07",[1088,1089,1091,711,1090],[1093,1094,1096,716,1095],[1484,1490,1491,1492,1493,1494,1495,1496],"langfuse vs langsmith vergleich","llm-tracing werkzeug","langfuse selbst hosten","langfuse preise","opentelemetry llm-traces","llm-kosten tracking","prompt-versionierung",[1498,1500,1501,1504],{"name":259,"url":1499},"https:\u002F\u002Flangfuse.com",{"name":1091,"url":1110},{"name":1502,"url":1503},"ClickHouse","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FClickHouse",{"name":1112,"url":1113},[1506,1509,1512,1515,1518,1521,1524,1527],{"title":1507,"url":1508},"Langfuse documentation: observability and application tracing","https:\u002F\u002Flangfuse.com\u002Fdocs\u002Fobservability\u002Foverview",{"title":1510,"url":1511},"Langfuse documentation: get started with tracing","https:\u002F\u002Flangfuse.com\u002Fdocs\u002Fobservability\u002Fget-started",{"title":1513,"url":1514},"Langfuse pricing: cloud plans, billable units and worked examples","https:\u002F\u002Flangfuse.com\u002Fpricing",{"title":1516,"url":1517},"Langfuse pricing: self-hosted plans and the feature comparison","https:\u002F\u002Flangfuse.com\u002Fpricing-self-host",{"title":1519,"url":1520},"Self-host Langfuse: deployment options, containers and storage services","https:\u002F\u002Flangfuse.com\u002Fself-hosting",{"title":1522,"url":1523},"Langfuse changelog: v4 is live (17 August 2026)","https:\u002F\u002Flangfuse.com\u002Fchangelog\u002F2026-08-17-langfuse-v4",{"title":1525,"url":1526},"Langfuse blog: Langfuse joins ClickHouse (16 January 2026)","https:\u002F\u002Flangfuse.com\u002Fblog\u002Fjoining-clickhouse",{"title":1528,"url":1180},"GitHub: langfuse\u002Flangfuse, the platform repository","\u002Fimages\u002Fblog\u002Flangfuse\u002Fcover.webp","\u002Fimages\u002Fblog\u002Flangfuse\u002Fog.jpg",[72,73,74],"Langfuse-Test: Tracing, Prompts und Evals selbst gehostet","Langfuse im Test: LLM-Traces, Prompt-Versionen und Experimente auf einer MIT-lizenzierten Plattform. Was Selbsthosting kostet und wo Langfuse verliert.","Langfuse: Eine Fußspur im Garten eines Hauses führt zu einem Notizbuch mit Lupe.",{"slug":1536,"published":1537,"updated":1195,"minutes":1538,"category":8,"tags":1539,"tagPages":1541,"keywords":1543,"about":1554,"sources":1558,"cover":1608,"og":1609,"expertise":70,"locales":1610,"lang":73,"title":1611,"description":1612,"coverAlt":1613},"agent-observability-opentelemetry","2026-08-06",12,[1091,1088,1540,1089,258],"AI agents",[1096,1093,1542,1094],"ai-agents",[1544,1545,1546,1547,1548,1549,1550,1551,1552,1553],"llm-agenten observability opentelemetry","genai-semantikkonventionen opentelemetry","llm-agenten tracen anleitung","gen_ai semantic conventions attribute","llm token-verbrauch kosten metriken","sampling bei llm-tracing","pii in llm-traces","langfuse vs arize phoenix vs datadog vergleich","ki-agenten tracing evals","opentelemetry llm tracing tutorial",[1555,1556,1557],{"name":1091,"url":1110},{"name":1301,"url":1113},{"name":572,"url":573},[1559,1562,1565,1568,1571,1574,1577,1580,1583,1586,1589,1592,1595,1597,1599,1602,1605],{"title":1560,"url":1561},"OpenTelemetry: GenAI semantic conventions repository (open-telemetry\u002Fsemantic-conventions-genai)","https:\u002F\u002Fgithub.com\u002Fopen-telemetry\u002Fsemantic-conventions-genai",{"title":1563,"url":1564},"GenAI conventions: overview (status Development)","https:\u002F\u002Fgithub.com\u002Fopen-telemetry\u002Fsemantic-conventions-genai\u002Fblob\u002Fmain\u002Fdocs\u002Fgen-ai\u002FREADME.md",{"title":1566,"url":1567},"GenAI conventions: model spans, execute_tool and content capture","https:\u002F\u002Fgithub.com\u002Fopen-telemetry\u002Fsemantic-conventions-genai\u002Fblob\u002Fmain\u002Fdocs\u002Fgen-ai\u002Fgen-ai-spans.md",{"title":1569,"url":1570},"GenAI conventions: agent spans","https:\u002F\u002Fgithub.com\u002Fopen-telemetry\u002Fsemantic-conventions-genai\u002Fblob\u002Fmain\u002Fdocs\u002Fgen-ai\u002Fgen-ai-agent-spans.md",{"title":1572,"url":1573},"GenAI conventions: metrics","https:\u002F\u002Fgithub.com\u002Fopen-telemetry\u002Fsemantic-conventions-genai\u002Fblob\u002Fmain\u002Fdocs\u002Fgen-ai\u002Fgen-ai-metrics.md",{"title":1575,"url":1576},"GenAI conventions: inference token metrics","https:\u002F\u002Fgithub.com\u002Fopen-telemetry\u002Fsemantic-conventions-genai\u002Fblob\u002Fmain\u002Fdocs\u002Fgen-ai\u002Fgen-ai-token-metrics.md",{"title":1578,"url":1579},"GenAI conventions: events (gen_ai.evaluation.result)","https:\u002F\u002Fgithub.com\u002Fopen-telemetry\u002Fsemantic-conventions-genai\u002Fblob\u002Fmain\u002Fdocs\u002Fgen-ai\u002Fgen-ai-events.md",{"title":1581,"url":1582},"GenAI conventions: Model Context Protocol","https:\u002F\u002Fgithub.com\u002Fopen-telemetry\u002Fsemantic-conventions-genai\u002Fblob\u002Fmain\u002Fdocs\u002Fgen-ai\u002Fmcp.md",{"title":1584,"url":1585},"OpenTelemetry docs: GenAI conventions moved notice","https:\u002F\u002Fopentelemetry.io\u002Fdocs\u002Fspecs\u002Fsemconv\u002Fgen-ai\u002F",{"title":1587,"url":1588},"OpenTelemetry docs: Sampling","https:\u002F\u002Fopentelemetry.io\u002Fdocs\u002Fconcepts\u002Fsampling\u002F",{"title":1590,"url":1591},"John Hodge: The state of the OpenTelemetry GenAI semantic conventions (July 2026)","https:\u002F\u002Fjohn-hodge.com\u002Fblog\u002Fopentelemetry-genai-semantic-conventions\u002F",{"title":1593,"url":1594},"Langfuse docs: OpenTelemetry integration","https:\u002F\u002Flangfuse.com\u002Fdocs\u002Fopentelemetry\u002Fget-started",{"title":1596,"url":1180},"Langfuse repository and licence",{"title":1598,"url":1183},"Arize Phoenix repository",{"title":1600,"url":1601},"Arize OpenInference repository","https:\u002F\u002Fgithub.com\u002FArize-ai\u002Fopeninference",{"title":1603,"url":1604},"Datadog docs: OpenTelemetry instrumentation for LLM Observability","https:\u002F\u002Fdocs.datadoghq.com\u002Fllm_observability\u002Finstrumentation\u002Fotel_instrumentation\u002F",{"title":1606,"url":1607},"Honeycomb docs: Send data with OpenTelemetry","https:\u002F\u002Fdocs.honeycomb.io\u002Fsend-data\u002Fopentelemetry\u002F","\u002Fimages\u002Fblog\u002Fagent-observability-opentelemetry\u002Fcover.webp","\u002Fimages\u002Fblog\u002Fagent-observability-opentelemetry\u002Fog.jpg",[72,73,74],"Observability für LLM-Agenten mit OpenTelemetry: Traces, Tokens, PII und Evals","So tracest du LLM-Agenten mit OpenTelemetry: GenAI Semantic Conventions und ihr Status, Span-Baum, Token-Metriken, Sampling, PII, Evals und Tool-Optionen.","Agent-Observability: Ein Stethoskop liegt auf einem kleinen Roboter neben einem Monitor mit ruhiger Wellenlinie.",{"slug":1615,"published":1616,"updated":1617,"minutes":184,"category":8,"tags":1618,"tagPages":1622,"keywords":1623,"about":1631,"sources":1634,"cover":1649,"og":1650,"expertise":70,"locales":1651,"lang":73,"title":1652,"description":1653,"coverAlt":1654},"openrouter","2026-07-23","2026-08-01",[190,86,1619,1620,1621],"Fallbacks","OpenAI-compatible","Pay per token",[],[1615,1624,1625,1626,1627,1628,1629,1630],"openrouter vs litellm vergleich","llm-gateway übersicht","openrouter preise","openai-kompatible api-gateway","llm fallback-routing","multi-modell api-gateway","byok llm-routing",[1632],{"name":1437,"url":1633},"https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FOpenRouter",[1635,1638,1639,1641,1644,1647],{"title":1636,"url":1637},"OpenRouter documentation: quickstart","https:\u002F\u002Fopenrouter.ai\u002Fdocs\u002Fquickstart",{"title":1370,"url":1371},{"title":1640,"url":241},"OpenRouter documentation: model fallbacks",{"title":1642,"url":1643},"OpenRouter documentation: provider routing","https:\u002F\u002Fopenrouter.ai\u002Fdocs\u002Fguides\u002Frouting\u002Fprovider-selection",{"title":1645,"url":1646},"OpenRouter documentation index","https:\u002F\u002Fopenrouter.ai\u002Fdocs\u002Fllms.txt",{"title":1648,"url":1633},"Wikipedia: OpenRouter","\u002Fimages\u002Fblog\u002Fopenrouter\u002Fcover.webp","\u002Fimages\u002Fblog\u002Fopenrouter\u002Fog.jpg",[72,73,74],"OpenRouter: ein API-Schlüssel vor jedem Modell im Katalog","OpenRouter im Test: 500+ Modelle von über 80 Anbietern hinter einem OpenAI-kompatiblen Endpunkt, mit Fallbacks und Weiterverrechnung zum Listenpreis.","OpenRouter: Ein Schlüssel am Ring, von dem Linien zu einer Reihe verschiedener Türen führen.",{"slug":1656,"published":1657,"minutes":184,"category":8,"tags":1658,"tagPages":1661,"keywords":1662,"about":1669,"sources":1675,"cover":1697,"og":1698,"expertise":70,"locales":1699,"lang":73,"title":1700,"description":1701,"coverAlt":1702},"braintrust","2026-07-07",[1659,1301,819,1660,1089],"Evaluations","CI gates",[1094],[1656,1663,1664,1665,1666,1667,1668],"braintrust preise","braintrust eval anleitung","autoevals bibliothek","braintrust vs langfuse vergleich","llm-evaluation plattform","eval-getriebene entwicklung",[1670,1673,1674],{"name":1671,"url":1672},"Braintrust","https:\u002F\u002Fwww.braintrust.dev",{"name":280,"url":281},{"name":1112,"url":1113},[1676,1678,1681,1684,1687,1690,1693,1695],{"title":1677,"url":896},"Braintrust pricing: plans, credits and usage rates",{"title":1679,"url":1680},"Braintrust documentation: plans and limits","https:\u002F\u002Fwww.braintrust.dev\u002Fdocs\u002Fplans-and-limits",{"title":1682,"url":1683},"Braintrust documentation: evaluation quickstart","https:\u002F\u002Fwww.braintrust.dev\u002Fdocs\u002Fevaluation-quickstart",{"title":1685,"url":1686},"Braintrust documentation: get started","https:\u002F\u002Fwww.braintrust.dev\u002Fdocs",{"title":1688,"url":1689},"GitHub: Braintrust organisation repositories","https:\u002F\u002Fgithub.com\u002Forgs\u002Fbraintrustdata\u002Frepositories",{"title":1691,"url":1692},"Arize Phoenix documentation: self-hosting","https:\u002F\u002Farize.com\u002Fdocs\u002Fphoenix\u002Fself-hosting",{"title":1694,"url":1514},"Langfuse pricing: cloud plans and billable units",{"title":1696,"url":1186},"LangChain pricing: LangSmith plans","\u002Fimages\u002Fblog\u002Fbraintrust\u002Fcover.webp","\u002Fimages\u002Fblog\u002Fbraintrust\u002Fog.jpg",[72,73,74],"Braintrust im Test: Eval-first Observability mit hartem Zähler","Braintrust im Test: Datasets und geprüfte Experimente aus Produktions-Traces. Was Starter und Pro kosten und wo Phoenix, Langfuse und LangSmith gewinnen.","Braintrust: Drei unterschiedlich gefüllte Reagenzgläser stehen neben einem großen runden Zähler.",{"slug":1704,"published":1657,"updated":1486,"minutes":184,"category":8,"tags":1705,"tagPages":1708,"keywords":1709,"about":1717,"sources":1721,"cover":1745,"og":1746,"expertise":70,"locales":1747,"lang":73,"title":1748,"description":1749,"coverAlt":1750},"promptfoo",[817,820,1706,1707],"Prompt testing","CI\u002FCD",[],[1704,1710,1711,1712,1713,1714,1715,1716],"promptfoo vs langsmith vergleich","llm-evaluierung werkzeug","promptfoo red teaming anleitung","promptfooconfig.yaml beispiel","llm-evals in ci","promptfoo preise","open-source llm-tests",[1718],{"name":1719,"url":1720},"Promptfoo","https:\u002F\u002Fwww.promptfoo.dev",[1722,1725,1728,1731,1734,1737,1740,1742],{"title":1723,"url":1724},"Promptfoo documentation: intro","https:\u002F\u002Fwww.promptfoo.dev\u002Fdocs\u002Fintro\u002F",{"title":1726,"url":1727},"Promptfoo pricing: plan comparison","https:\u002F\u002Fwww.promptfoo.dev\u002Fpricing",{"title":1729,"url":1730},"Promptfoo documentation: getting started","https:\u002F\u002Fwww.promptfoo.dev\u002Fdocs\u002Fgetting-started\u002F",{"title":1732,"url":1733},"Promptfoo documentation: command line usage (exit codes)","https:\u002F\u002Fwww.promptfoo.dev\u002Fdocs\u002Fusage\u002Fcommand-line\u002F",{"title":1735,"url":1736},"Promptfoo documentation: red team plugins","https:\u002F\u002Fwww.promptfoo.dev\u002Fdocs\u002Fred-team\u002Fplugins\u002F",{"title":1738,"url":1739},"Promptfoo documentation: assertions and metrics","https:\u002F\u002Fwww.promptfoo.dev\u002Fdocs\u002Fconfiguration\u002Fexpected-outputs\u002F",{"title":1741,"url":893},"Promptfoo repository on GitHub (MIT licence)",{"title":1743,"url":1744},"Promptfoo blog: Promptfoo is joining OpenAI (9 March 2026)","https:\u002F\u002Fwww.promptfoo.dev\u002Fblog\u002Fpromptfoo-joining-openai\u002F","\u002Fimages\u002Fblog\u002Fpromptfoo\u002Fcover.webp","\u002Fimages\u002Fblog\u002Fpromptfoo\u002Fog.jpg",[72,73,74],"Promptfoo: LLM-Evals und Red Teaming aus einer YAML-Datei","Promptfoo 2026: Evals und Red Teaming unter MIT, seit März 2026 im Besitz von OpenAI. Was es gut kann, wo der YAML-Ansatz bricht und was die Stufen kosten.","Promptfoo: Ein Klemmbrett vor einer Reihe Zielscheiben, ein Pfeil daneben.",{"slug":1752,"published":1753,"updated":1195,"minutes":184,"category":8,"tags":1754,"tagPages":1756,"keywords":1757,"about":1765,"sources":1770,"cover":1793,"og":1794,"expertise":70,"locales":1795,"lang":73,"title":1796,"description":1797,"coverAlt":1798},"arize-phoenix","2026-07-03",[1088,1089,1091,1755,1090],"Self-hosted",[1093,1094,1096,1095],[1758,1759,1760,1761,1762,1763,1764],"arize phoenix","arize phoenix selbst hosten","arize phoenix docker anleitung","llm-tracing opentelemetry","phoenix vs langfuse vergleich","openinference instrumentation","llm-observability tools vergleich",[1766,1768,1769],{"name":1767,"url":1183},"Arize Phoenix",{"name":1091,"url":1110},{"name":1112,"url":1113},[1771,1774,1776,1779,1782,1785,1788,1791,1792],{"title":1772,"url":1773},"Arize Phoenix: open-source AI observability and evaluation","https:\u002F\u002Fphoenix.arize.com",{"title":1775,"url":1692},"Phoenix documentation: self-hosting",{"title":1777,"url":1778},"Phoenix documentation: licence, Elastic License 2.0","https:\u002F\u002Farize.com\u002Fdocs\u002Fphoenix\u002Fself-hosting\u002Flicense",{"title":1780,"url":1781},"Phoenix documentation: architecture, storage and scaling","https:\u002F\u002Farize.com\u002Fdocs\u002Fphoenix\u002Fself-hosting\u002Farchitecture",{"title":1783,"url":1784},"Phoenix documentation: setup tracing","https:\u002F\u002Farize.com\u002Fdocs\u002Fphoenix\u002Ftracing\u002Fhow-to-tracing\u002Fsetup-tracing",{"title":1786,"url":1787},"Phoenix documentation: OpenTelemetry SDK setup","https:\u002F\u002Farize.com\u002Fdocs\u002Fphoenix\u002Ftracing\u002Fhow-to-tracing\u002Fsetup-tracing\u002Fsetup-using-phoenix-otel",{"title":1789,"url":1790},"Arize pricing: AX Free, AX Pro and AX Enterprise","https:\u002F\u002Farize.com\u002Fpricing\u002F",{"title":1694,"url":1514},{"title":1696,"url":1186},"\u002Fimages\u002Fblog\u002Farize-phoenix\u002Fcover.webp","\u002Fimages\u002Fblog\u002Farize-phoenix\u002Fog.jpg",[72,73,74],"Arize Phoenix im Test: LLM-Tracing und Evals auf eigener Infrastruktur","Arize Phoenix im Test: ELv2-lizenzierter Tracing- und Eval-Server, Betriebskosten und wo Langfuse, Braintrust und LangSmith besser sind.","Arize Phoenix: Ein Phönix steigt über einem kleinen Haus auf, auf dessen Dach eine Lupe liegt.",{"slug":1800,"published":1753,"minutes":184,"category":8,"tags":1801,"tagPages":1805,"keywords":1806,"about":1814,"sources":1818,"cover":1846,"og":1847,"expertise":70,"locales":1848,"lang":73,"title":1849,"description":1850,"coverAlt":1851},"helicone",[1088,1091,1802,1803,1804],"Cost tracking","Proxy","Gateway",[1093,1096],[1800,1807,1808,1809,1810,1811,1812,1813],"helicone ki-observability","helicone vs langfuse vergleich","llm-kostentracking tool","open-source llm monitoring","llm-gateway proxy","llm observability selbst hosten","helicone cache anleitung",[1815,1816],{"name":1091,"url":1110},{"name":1817,"url":1116},"Apache License 2.0",[1819,1822,1825,1828,1831,1834,1837,1840,1843],{"title":1820,"url":1821},"Helicone quickstart","https:\u002F\u002Fdocs.helicone.ai\u002Fgetting-started\u002Fquick-start",{"title":1823,"url":1824},"Helicone AI Gateway overview","https:\u002F\u002Fdocs.helicone.ai\u002Fgateway\u002Foverview",{"title":1826,"url":1827},"Helicone: latency impact and benchmark","https:\u002F\u002Fdocs.helicone.ai\u002Freferences\u002Flatency-affect",{"title":1829,"url":1830},"Helicone: proxy versus async integration","https:\u002F\u002Fdocs.helicone.ai\u002Freferences\u002Fproxy-vs-async",{"title":1832,"url":1833},"Helicone: LLM caching","https:\u002F\u002Fdocs.helicone.ai\u002Ffeatures\u002Fadvanced-usage\u002Fcaching",{"title":1835,"url":1836},"Helicone: self-hosting with Docker","https:\u002F\u002Fdocs.helicone.ai\u002Fgetting-started\u002Fself-deploy-docker",{"title":1838,"url":1839},"Helicone: how we calculate cost","https:\u002F\u002Fdocs.helicone.ai\u002Ffaq\u002Fhow-we-calculate-cost",{"title":1841,"url":1842},"Helicone pricing","https:\u002F\u002Fwww.helicone.ai\u002Fpricing",{"title":1844,"url":1845},"Helicone repository on GitHub","https:\u002F\u002Fgithub.com\u002FHelicone\u002Fhelicone","\u002Fimages\u002Fblog\u002Fhelicone\u002Fcover.webp","\u002Fimages\u002Fblog\u002Fhelicone\u002Fog.jpg",[72,73,74],"Helicone: Observability, die im Request-Pfad sitzt","Helicone im Test: Apache-2.0-Plattform für LLM-Gateway und Observability. Was der Proxy bringt, was er kostet und wie der Eigenbetrieb aussieht.","Helicone: Ein Umschlag auf Rädern fährt durch eine Mautstation mit Kamera.",{"slug":1853,"published":1854,"minutes":7,"category":8,"tags":1855,"tagPages":1861,"keywords":1862,"about":1870,"sources":1873,"cover":1901,"og":1902,"expertise":70,"locales":1903,"lang":73,"title":1904,"description":1905,"coverAlt":1906},"vllm","2026-06-30",[1856,1857,1858,1859,1860],"Self-hosted inference","OpenAI API","PagedAttention","GPU serving","LLM runtime",[],[1853,1863,1864,1865,1866,1867,1868,1869],"vllm test","vllm vs sglang vergleich","vllm vs tensorrt-llm vergleich","llm-server selbst hosten","openai-kompatibler api-server","llm inferenz durchsatz","pagedattention erklärt",[1871],{"name":713,"url":1872},"https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FvLLM",[1874,1877,1880,1883,1886,1889,1892,1895,1898],{"title":1875,"url":1876},"vLLM documentation","https:\u002F\u002Fdocs.vllm.ai\u002Fen\u002Flatest\u002F",{"title":1878,"url":1879},"Optimization and tuning for the V1 engine","https:\u002F\u002Fdocs.vllm.ai\u002Fen\u002Fstable\u002Fconfiguration\u002Foptimization.html",{"title":1881,"url":1882},"Architecture overview and the V1 process layout","https:\u002F\u002Fdocs.vllm.ai\u002Fen\u002Fstable\u002Fdesign\u002Farch_overview.html",{"title":1884,"url":1885},"Metrics design","https:\u002F\u002Fdocs.vllm.ai\u002Fen\u002Fstable\u002Fdesign\u002Fmetrics.html",{"title":1887,"url":1888},"Automatic prefix caching and its documented limits","https:\u002F\u002Fdocs.vllm.ai\u002Fen\u002Fstable\u002Ffeatures\u002Fautomatic_prefix_caching.html",{"title":1890,"url":1891},"Online serving and the HTTP API surface","https:\u002F\u002Fdocs.vllm.ai\u002Fen\u002Fstable\u002Fserving\u002Fonline_serving.html",{"title":1893,"url":1894},"Inside vLLM: anatomy of a high-throughput inference system","https:\u002F\u002Fvllm.ai\u002Fblog\u002F2025-09-05-anatomy-of-vllm",{"title":1896,"url":1897},"vLLM: easy, fast and cheap LLM serving with PagedAttention","https:\u002F\u002Fblog.vllm.ai\u002F2023\u002F06\u002F20\u002Fvllm.html",{"title":1899,"url":1900},"Release history","https:\u002F\u002Fgithub.com\u002Fvllm-project\u002Fvllm\u002Freleases","\u002Fimages\u002Fblog\u002Fvllm\u002Fcover.webp","\u002Fimages\u002Fblog\u002Fvllm\u002Fog.jpg",[72,73,74],"vLLM geprüft: Inferenz auf eigener Hardware","vLLM: aus einem Hugging-Face-Checkpoint ein OpenAI-kompatibler Server. Was PagedAttention und kontinuierliches Batching bringen, was der Betrieb kostet.","vLLM: Ein großer Ofen backt viele Bleche Brot gleichzeitig, daneben ein Brotschieber.",{"slug":1908,"published":1909,"updated":1910,"minutes":315,"category":8,"tags":1911,"tagPages":1912,"keywords":1913,"about":1922,"sources":1927,"cover":1935,"og":1936,"expertise":70,"locales":1937,"lang":73,"title":1938,"description":1939,"coverAlt":1940},"llm-cost-latency-prompt-caching-routing","2026-06-01","2026-09-22",[319,86,133,665],[],[1914,1915,1916,1917,1918,1919,1920,1921],"prompt-caching","llm-kosten senken","llm latenz senken","modell-routing","batch-api kosten senken","llm-kosten pro anfrage","günstigeres llm-modell","cache-trefferquote llm",[1923,1924],{"name":341,"url":342},{"name":1925,"url":1926},"Latency (engineering)","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FLatency_(engineering)",[1928,1931,1934],{"title":1929,"url":1930},"Claude API docs: Prompt caching","https:\u002F\u002Fplatform.claude.com\u002Fdocs\u002Fen\u002Fdocs\u002Fbuild-with-claude\u002Fprompt-caching",{"title":1932,"url":1933},"OpenAI API docs: Prompt caching","https:\u002F\u002Fdevelopers.openai.com\u002Fapi\u002Fdocs\u002Fguides\u002Fprompt-caching",{"title":1423,"url":118},"\u002Fimages\u002Fblog\u002Fllm-cost-latency-prompt-caching-routing\u002Fcover.webp","\u002Fimages\u002Fblog\u002Fllm-cost-latency-prompt-caching-routing\u002Fog.jpg",[72,73,74],"Prompt-Caching und Modell-Routing: LLM-Kosten und Latenz senken","LLM-Kosten und Latenz senken: Prompt-Caching, Routing auf das kleinere Modell und Batch-APIs sind die Hebel im Produktivbetrieb. So nutzt du jeden davon.","Prompt-Caching und Routing: Eine Weiche teilt das Gleis zu einem kleinen und einem großen Bahnhof, daneben Uhr und Münze.",{"slug":1942,"published":1943,"updated":1944,"minutes":184,"category":8,"tags":1945,"tagPages":1947,"keywords":1948,"about":1956,"sources":1962,"cover":1983,"og":1984,"expertise":70,"locales":1985,"lang":73,"title":1986,"description":1987,"coverAlt":1988},"langsmith","2026-05-25","2026-09-14",[1089,612,1091,1946,187],"Datasets",[1094,1096],[1942,1949,1950,1951,1952,1953,1954,1955],"langsmith vs langfuse vergleich","llm observability werkzeuge","llm-tracing kosten","langsmith preise","langsmith opentelemetry","llm-evaluation datasets","phoenix vs langsmith vergleich",[1957,1960,1961],{"name":1958,"url":1959},"LangSmith","https:\u002F\u002Fwww.langchain.com\u002Flangsmith",{"name":1091,"url":1110},{"name":1502,"url":1503},[1963,1965,1968,1971,1974,1977,1980],{"title":1964,"url":1186},"LangSmith pricing",{"title":1966,"url":1967},"LangSmith documentation: administration overview","https:\u002F\u002Fdocs.langchain.com\u002Flangsmith\u002Fadministration-overview",{"title":1969,"url":1970},"LangSmith documentation: custom instrumentation","https:\u002F\u002Fdocs.langchain.com\u002Flangsmith\u002Fannotate-code",{"title":1972,"url":1973},"LangSmith documentation: trace with OpenTelemetry","https:\u002F\u002Fdocs.langchain.com\u002Flangsmith\u002Ftrace-with-opentelemetry",{"title":1975,"url":1976},"LangSmith documentation: evaluation concepts","https:\u002F\u002Fdocs.langchain.com\u002Flangsmith\u002Fevaluation-concepts",{"title":1978,"url":1979},"LangSmith documentation: self-hosted deployment","https:\u002F\u002Fdocs.langchain.com\u002Flangsmith\u002Fself-hosted",{"title":1981,"url":1982},"LangChain: introducing end-to-end OpenTelemetry support in LangSmith","https:\u002F\u002Fwww.langchain.com\u002Fblog\u002Fend-to-end-opentelemetry-langsmith","\u002Fimages\u002Fblog\u002Flangsmith\u002Fcover.webp","\u002Fimages\u002Fblog\u002Flangsmith\u002Fog.jpg",[72,73,74],"LangSmith: aus einem Observability-Produkt wurde eine Agent-Plattform","LangSmith im Test: Abrechnung pro Trace, 14 und 180 Tage Retention, OpenTelemetry-Ingestion, Evals und der Schwenk zur Agent-Runtime.","LangSmith: Ein Amboss mit Hammer steht neben einem Teleskop und einer wachsenden Pflanze.",{"slug":1990,"published":1991,"updated":1992,"minutes":184,"category":8,"tags":1993,"tagPages":1995,"keywords":1997,"about":2004,"sources":2011,"cover":2030,"og":2031,"expertise":70,"locales":2032,"lang":73,"title":2033,"description":2034,"coverAlt":2035},"litellm","2026-05-20","2026-07-09",[190,1620,1302,1300,1994],"MCP",[1996],"mcp",[1990,1998,1308,1999,2000,2001,2002,2003],"litellm vs portkey vergleich","openai-kompatibler proxy","litellm proxy konfiguration yaml","llm-ausgaben pro team","llm-gateway selbst hosten","llm budgetlimit",[2005,2008],{"name":2006,"url":2007},"LiteLLM","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FLiteLLM",{"name":2009,"url":2010},"Model Context Protocol","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FModel_Context_Protocol",[2012,2015,2018,2020,2023,2026,2029],{"title":2013,"url":2014},"LiteLLM documentation: getting started","https:\u002F\u002Fdocs.litellm.ai\u002Fdocs\u002F",{"title":2016,"url":2017},"LiteLLM documentation: virtual keys","https:\u002F\u002Fdocs.litellm.ai\u002Fdocs\u002Fproxy\u002Fvirtual_keys",{"title":2019,"url":235},"LiteLLM documentation: router and load balancing",{"title":2021,"url":2022},"LiteLLM documentation: benchmarks","https:\u002F\u002Fdocs.litellm.ai\u002Fdocs\u002Fbenchmarks",{"title":2024,"url":2025},"LiteLLM documentation: supported endpoints","https:\u002F\u002Fdocs.litellm.ai\u002Fdocs\u002Fsupported_endpoints",{"title":2027,"url":2028},"LiteLLM documentation: Enterprise","https:\u002F\u002Fdocs.litellm.ai\u002Fdocs\u002Fenterprise",{"title":1367,"url":1368},"\u002Fimages\u002Fblog\u002Flitellm\u002Fcover.webp","\u002Fimages\u002Fblog\u002Flitellm\u002Fog.jpg",[72,73,74],"LiteLLM: das OpenAI-kompatible Gateway, das die meisten Plattformteams betreiben","LiteLLM im Test: das MIT-lizenzierte OpenAI-kompatible Gateway, das die meisten Plattformteams vor ihre Anbieter setzen. Was es gut macht und wo es kippt.","LiteLLM: Ein Universal-Reiseadapter führt mehrere verschiedene Stecker zu einem Kabel zusammen.",{"slug":2037,"published":1991,"minutes":184,"category":8,"tags":2038,"tagPages":2043,"keywords":2044,"about":2052,"sources":2055,"cover":2073,"og":2074,"expertise":70,"locales":2075,"lang":73,"title":2076,"description":2077,"coverAlt":2078},"ragas",[2039,555,2040,2041,2042],"RAG evaluation","Test sets","CI quality","Ragas",[],[2037,2045,2046,2047,2048,2049,2050,2051],"ragas anleitung","ragas vs deepeval vergleich","llm-evaluation metriken","faithfulness context precision erklärt","rag-evaluierungs-framework","llm-as-a-judge ragas","ragas kosten",[2053],{"name":2042,"url":2054},"https:\u002F\u002Fwww.ragas.io\u002F",[2056,2059,2062,2065,2068,2071],{"title":2057,"url":2058},"Ragas documentation: introduction","https:\u002F\u002Fdocs.ragas.io\u002Fen\u002Fstable\u002F",{"title":2060,"url":2061},"Ragas documentation: quick start","https:\u002F\u002Fdocs.ragas.io\u002Fen\u002Fstable\u002Fgetstarted\u002Fquickstart\u002F",{"title":2063,"url":2064},"Ragas documentation: list of available metrics","https:\u002F\u002Fdocs.ragas.io\u002Fen\u002Fstable\u002Fconcepts\u002Fmetrics\u002Favailable_metrics\u002F",{"title":2066,"url":2067},"Ragas on GitHub","https:\u002F\u002Fgithub.com\u002Fvibrantlabsai\u002Fragas",{"title":2069,"url":2070},"Ragas on PyPI","https:\u002F\u002Fpypi.org\u002Fproject\u002Fragas\u002F",{"title":2072,"url":2054},"Ragas website","\u002Fimages\u002Fblog\u002Fragas\u002Fcover.webp","\u002Fimages\u002Fblog\u002Fragas\u002Fog.jpg",[72,73,74],"Ragas-Review: der Standard für RAG-Evaluation","Ragas im Review: RAG- und Agenten-Pipelines mit Faithfulness, Context Precision und Context Recall bewerten. Apache-2.0, das Judge-Modell kostet extra.","Ragas: Ein Zeugnis mit Kästchen und Stern neben Büchern und einem Bleistift.",{"slug":2080,"published":2081,"updated":1432,"minutes":253,"category":8,"tags":2082,"tagPages":2085,"keywords":2086,"about":2095,"sources":2098,"cover":2115,"og":2116,"expertise":70,"locales":2117,"lang":73,"title":2118,"description":2119,"coverAlt":2120},"llm-evals-for-product-features","2026-05-15",[612,555,2083,2084],"Error analysis","CI",[],[2087,2088,2089,2090,2091,2092,2093,2094],"llm-evals produktfeatures","ki-evals","llm als richter","eval-driven development","pass^k vs pass@k vergleich","agenten-evaluation harness","regressions-eval-suite","fehleranalyse llm",[2096,2097],{"name":341,"url":342},{"name":575,"url":576},[2099,2101,2104,2107,2110,2112],{"title":2100,"url":581},"Anthropic: Demystifying evals for AI agents (2026)",{"title":2102,"url":2103},"Hamel Husain: LLM evals FAQ (updated September 2026)","https:\u002F\u002Fhamel.dev\u002Fblog\u002Fposts\u002Fevals-faq\u002F",{"title":2105,"url":2106},"Shankar et al., Who Validates the Validators? (2024)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2404.12272",{"title":2108,"url":2109},"Zheng et al., Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena (2023)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2306.05685",{"title":2111,"url":584},"Yao et al., tau-bench: A Benchmark for Tool-Agent-User Interaction (2024)",{"title":2113,"url":2114},"Anthropic: Quantifying infrastructure noise in agentic coding evals (2026)","https:\u002F\u002Fwww.anthropic.com\u002Fengineering\u002Finfrastructure-noise","\u002Fimages\u002Fblog\u002Fllm-evals-for-product-features\u002Fcover.webp","\u002Fimages\u002Fblog\u002Fllm-evals-for-product-features\u002Fog.jpg",[72,73,74],"LLM-Evals für Produktfeatures: von handgelesenen Traces zum CI-Gate","LLM-Evals machen aus dem Bauchgefühl eine Testsuite: Fehleranalyse auf echten Traces, Grader-Wahl, ein validierter LLM-as-judge, pass^k und ein CI-Gate.","LLM-Evals für Produktfeatures: Kisten auf einem Band passieren ein Prüftor, eine Person liest mit Lupe.",{"slug":2122,"published":2123,"updated":2124,"minutes":1538,"category":8,"tags":2125,"tagPages":2129,"keywords":2131,"about":2142,"sources":2148,"cover":2215,"og":2216,"expertise":70,"locales":2217,"lang":73,"title":2218,"description":2219,"coverAlt":2220},"fine-tuning-vs-rag-vs-prompting","2026-05-12","2026-10-02",[2126,2127,2128,374],"Fine-tuning","RAG","Prompting",[2130],"rag",[2132,2133,2134,2135,2136,2137,2138,2139,2140,2141],"fine-tuning oder rag","prompting vs rag vs fine-tuning vergleich","wann llm feintunen","rag oder feintuning","sft vs dpo vs rft","reinforcement fine-tuning anbieter","llm-destillation","openai fine-tuning abschaltung","fine-tuning entscheidungsbaum","fine-tuning für wissen",[2143,2144,2147],{"name":930,"url":931},{"name":2145,"url":2146},"Retrieval-augmented generation","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FRetrieval-augmented_generation",{"name":394,"url":395},[2149,2152,2155,2158,2161,2164,2167,2170,2173,2176,2179,2182,2185,2188,2191,2194,2197,2200,2203,2206,2209,2212],{"title":2150,"url":2151},"OpenAI community: OpenAI self-serve fine-tuning availability (wind-down announcement)","https:\u002F\u002Fcommunity.openai.com\u002Ft\u002Fopenai-s-self-serve-fine-tuning-availability\u002F1380481",{"title":2153,"url":2154},"Tessl: OpenAI is shutting down self-serve fine-tuning","https:\u002F\u002Ftessl.io\u002Fblog\u002Fopenai-shutting-fine-tuning-signals-for-enterprise-ai",{"title":2156,"url":2157},"OpenAI API docs: Model optimization","https:\u002F\u002Fdevelopers.openai.com\u002Fapi\u002Fdocs\u002Fguides\u002Fmodel-optimization",{"title":2159,"url":2160},"OpenAI API docs: Supervised fine-tuning","https:\u002F\u002Fdevelopers.openai.com\u002Fapi\u002Fdocs\u002Fguides\u002Fsupervised-fine-tuning",{"title":2162,"url":2163},"OpenAI API docs: Direct preference optimization","https:\u002F\u002Fdevelopers.openai.com\u002Fapi\u002Fdocs\u002Fguides\u002Fdirect-preference-optimization",{"title":2165,"url":2166},"OpenAI API docs: Reinforcement fine-tuning","https:\u002F\u002Fdevelopers.openai.com\u002Fapi\u002Fdocs\u002Fguides\u002Freinforcement-fine-tuning",{"title":2168,"url":2169},"OpenAI API docs: Fine-tuning best practices","https:\u002F\u002Fdevelopers.openai.com\u002Fapi\u002Fdocs\u002Fguides\u002Ffine-tuning-best-practices",{"title":2171,"url":2172},"OpenAI Cookbook: Choosing between SFT, DPO and RFT","https:\u002F\u002Fdevelopers.openai.com\u002Fcookbook\u002Fexamples\u002Ffine_tuning_direct_preference_optimization_guide",{"title":2174,"url":2175},"Google Cloud: Introduction to tuning (Gemini)","https:\u002F\u002Fdocs.cloud.google.com\u002Fvertex-ai\u002Fgenerative-ai\u002Fdocs\u002Fmodels\u002Ftune-models",{"title":2177,"url":2178},"Google Cloud: About supervised fine-tuning for Gemini models","https:\u002F\u002Fdocs.cloud.google.com\u002Fvertex-ai\u002Fgenerative-ai\u002Fdocs\u002Fmodels\u002Fgemini-supervised-tuning",{"title":2180,"url":2181},"Google Cloud: About preference tuning for Gemini models","https:\u002F\u002Fdocs.cloud.google.com\u002Fvertex-ai\u002Fgenerative-ai\u002Fdocs\u002Fmodels\u002Fgemini-preference-tuning",{"title":2183,"url":2184},"Google Cloud: Reward functions for reinforcement learning fine-tuning","https:\u002F\u002Fdocs.cloud.google.com\u002Fgemini-enterprise-agent-platform\u002Fmodels\u002Ftuning\u002Freinforcement-tuning\u002Freinforcement-tuning-job\u002Freward-functions",{"title":2186,"url":2187},"Google Cloud: Supervised and distillation fine-tuning for open models","https:\u002F\u002Fdocs.cloud.google.com\u002Fvertex-ai\u002Fgenerative-ai\u002Fdocs\u002Fmodels\u002Fopen-model-tuning",{"title":2189,"url":2190},"AWS: Fine-tuning for Claude 3 Haiku in Amazon Bedrock is now generally available","https:\u002F\u002Faws.amazon.com\u002Fblogs\u002Faws\u002Ffine-tuning-for-anthropics-claude-3-haiku-model-in-amazon-bedrock-is-now-generally-available\u002F",{"title":2192,"url":2193},"AWS: Amazon Bedrock now supports reinforcement fine-tuning","https:\u002F\u002Faws.amazon.com\u002Fabout-aws\u002Fwhats-new\u002F2025\u002F12\u002Fbedrock-reinforcement-fine-tuning-66-base-models",{"title":2195,"url":2196},"AWS: Amazon Bedrock Model Distillation (preview announcement)","https:\u002F\u002Faws.amazon.com\u002Fblogs\u002Faws\u002Fbuild-faster-more-cost-efficient-highly-accurate-models-with-amazon-bedrock-model-distillation-preview\u002F",{"title":2198,"url":2199},"AWS: Bedrock reinforcement fine-tuning adds open-weight models","https:\u002F\u002Faws.amazon.com\u002Fabout-aws\u002Fwhats-new\u002F2026\u002F02\u002Famazon-bedrock-reinforcement-fine-tuning-openai",{"title":2201,"url":2202},"Microsoft Foundry blog: What is new in Foundry fine-tuning, April 2026","https:\u002F\u002Fdevblogs.microsoft.com\u002Ffoundry\u002Fwhats-new-in-foundry-finetune-april-2026\u002F",{"title":2204,"url":2205},"Microsoft: Announcing new fine-tuning models and techniques in Azure AI Foundry","https:\u002F\u002Fazure.microsoft.com\u002Fen-us\u002Fblog\u002Fannouncing-new-fine-tuning-models-and-techniques-in-azure-ai-foundry\u002F",{"title":2207,"url":2208},"Ovadia et al.: Fine-Tuning or Retrieval? Comparing Knowledge Injection in LLMs","https:\u002F\u002Farxiv.org\u002Fabs\u002F2312.05934",{"title":2210,"url":2211},"Gekhman et al.: Does Fine-Tuning LLMs on New Knowledge Encourage Hallucinations?","https:\u002F\u002Farxiv.org\u002Fabs\u002F2405.05904",{"title":2213,"url":2214},"Hu et al.: LoRA: Low-Rank Adaptation of Large Language Models","https:\u002F\u002Farxiv.org\u002Fabs\u002F2106.09685","\u002Fimages\u002Fblog\u002Ffine-tuning-vs-rag-vs-prompting\u002Fcover.webp","\u002Fimages\u002Fblog\u002Ffine-tuning-vs-rag-vs-prompting\u002Fog.jpg",[72,73,74],"Prompting vs. RAG vs. Fine-Tuning vs. Distillation: Entscheidungshilfe für 2026","Prompting, RAG, Fine-Tuning oder Distillation: was jeder Hebel ändert (Wissen, Verhalten, Format), was er kostet und wer 2026 SFT, DPO und RFT anbietet.","Prompting, RAG, Fine-Tuning: Ein Roboter mit Karte vor vier Türen mit Sprechblase, Büchern, Schraubenschlüssel und Trichter.",1793451238872]