BerriAI/litellm v1.76.2-nightly

on GitHub

What's Changed

• Logs page screen size fixed by @NANDINI-star in #14135
• Create Organization Tooltip added on Success by @NANDINI-star in #14132
• [BUG] XAI Cost Calculation Fix by @kankute-sameer in #14127
• Back to Keys should say Back to Logs by @NANDINI-star in #14134
• helm(chart): add optional PodDisruptionBudget for litellm proxy (#14062) by @iabhi4 in #14093
• update together models by @zainhas in #14087
• [Perf] LiteLLM Proxy: +400 RPS when using correct amount of CPU cores by @ishaan-jaff in #14153
• [Bug fix] Misclassified 500 error on invalid image_url in /chat/completions request by @ishaan-jaff in #14149
• openrouter: added gpt 4.1 model family by @mubashir1osmani in #14101
• [Feat] Allow using x-litellm-stream-timeout header for stream timeout in requests by @ishaan-jaff in #14147
• [Bug]: Gemini 2.5 Pro – schema validation fails with OpenAI-style type arrays in tools by @ishaan-jaff in #14154
• [Bug Fix] Gemini Tool Calling - fix gemini empty enum property by @ishaan-jaff in #14155
• GPT-5: Drop unsupported params by @LifeDJIK in #14146
• Prometheus missing metrics by @mubashir1osmani in #14139
• Add supported text field to anthropic citation response by @TomeHirata in #14126
• Fix token count error for gemini cli by @retanoj in #14133
• Braintrust - fix logging when OTEL is enabled + Gemini - add 'thoughtSignature' support via 'thinking_blocks' by @krrishdholakia in #14122
• VLLM - handle output parsing responses api output + Ollama - add unified 'thinking' param support (via reasoning_content) by @krrishdholakia in #14121
• [Feat]Add support for safety_identifier parameter in chat.completions.create by @kankute-sameer in #14174
• [Feature]: Support GPT-OSS models on vertex ai by @ishaan-jaff in #14184
• [Feature]: Add header support for spend_logs_metadata by @ishaan-jaff in #14186
• OTEL: Optional Metrics and Logs following semantic conventions by @keith-decker in #14179
• fix: Log page parameter passing error by @zhxlp in #14193
• Remove "/" or ":" from model name when being used as h11 header name by @kayoch1n in #14191
• Remove table filter on user info page by @NANDINI-star in #14169
• fix(oci): Handle assistant messages with both content and tool_calls in OCI provider (#14158) by @kutsushitaneko in #14171
• feat: added alert type to alert messate to slack for easier handling by @mjmendo in #14176
• Fix/remove deprecated cerebras gpt oss 20b by @HarshavardhanK in #14213
• [Feat] Support reasoning_effort in Groq by @eycjur in #14207
• Feat - add better SCIM debugging by @ishaan-jaff in #14221
• [Fix] SCIM - Bug fixes for handling SCIM Group Memberships by @ishaan-jaff in #14226

New Contributors

• @iabhi4 made their first contribution in #14093
• @zainhas made their first contribution in #14087
• @LifeDJIK made their first contribution in #14146
• @retanoj made their first contribution in #14133
• @zhxlp made their first contribution in #14193
• @kayoch1n made their first contribution in #14191
• @kutsushitaneko made their first contribution in #14171
• @mjmendo made their first contribution in #14176
• @HarshavardhanK made their first contribution in #14213
• @eycjur made their first contribution in #14207

Full Changelog: v1.76.1.rc.1...v1.76.2-nightly

Docker Run LiteLLM Proxy

docker run \
-e STORE_MODEL_IN_DB=True \
-p 4000:4000 \
ghcr.io/berriai/litellm:main-v1.76.2-nightly

Don't want to maintain your internal proxy? get in touch 🎉

Hosted Proxy Alpha: https://calendly.com/d/4mp-gd3-k5k/litellm-1-1-onboarding-chat

Load Test LiteLLM Proxy Results