What's Changed
- (feat) Milvus - search vector store support + (fix) Passthrough Endpoints - support multi-part form data on passthrough by @krrishdholakia in #16035
- (feat) Azure AI Vector Stores - support "virtual" indexes + create vector store on passthrough API by @krrishdholakia in #16160
- Add
custom_llm_providersupport for video endpoints (non-generation) by @Sameerlite in #16121 - fix: resolve memory accumulation caused by Pydantic 2.11+ deprecation warnings by @AlexsanderHamir in #16110
- add: minimum resource requirement for production by @AlexsanderHamir in #16146
- Remove minimum validation for cache control injection index by @steve-gore-snapdocs in #16149
- Fix empty assistant message handling in AWS Bedrock Converse API to prevent 400 Bad Request errors by @Sameerlite in #15850
- Add LLM provider response headers to Responses API by @Sameerlite in #16091
- Fix azure doesn't accept extra body param by @Sameerlite in #16116
- Add E2E Container API Support by @Sameerlite in #16136
- Add new model pricing for nemotron-nano-9b-v2 by @timbmg in #16120
- docs - v1.79.1 by @ishaan-jaff in #16163
- UI - Fix regression where Guardrail Entity Could not be selected and entity was not displayed by @ishaan-jaff in #16165
- Fix Anthropic token counting for VertexAI by @steve-gore-snapdocs in #16171
- [Feature] UI - Delete Team Member with friction by @yuneng-jiang in #16167
- [Feature] UI - Add Model Existing Credentials Improvement by @yuneng-jiang in #16166
- [Feature] UI - Guardrail Info Page Show PII Config by @yuneng-jiang in #16164
- strip base64 in s3 by @deepanshululla in #16157
- fix(ui_sso.py): support dot notation on ui sso by @krrishdholakia in #16135
- Support responses API streaming in langfuse otel by @eycjur in #16153
- fix: return the diarized transcript when it's required in the request by @gvioss in #16133
- build: Squashed commit of the following: by @krrishdholakia in #16176
- feat: change guardrail_information to list type by @uc4w6c in #16127
- Milvus - Passthrough API support - adds create + read vector store support via passthrough API's by @krrishdholakia in #16170
- Fix index field not populated in streaming mode with n>1 and tool calls by @Sameerlite in #15962
- Update perplexity cost tracking by @Sameerlite in #15743
- Fix image_config.aspect_ratio not working for gemini-2.5-flash-image by @Sameerlite in #15999
- [Feat] s3 logger, add support for ssl_verify when using minio logger by @ishaan-jaff in #16211
- [Patch] - Revert Noma Apply Guardrail implementation by @ishaan-jaff in #16214
- [Feat] /ocr - Add VertexAI OCR provider support + cost tracking by @ishaan-jaff in #16216
- fix: broken link on model_management.md by @AlexsanderHamir in #16217
- [Feat] Add Azure AI Doc Intelligence OCR by @ishaan-jaff in #16219
- Add Prometheus metric to track callback logging failures in S3 by @Sameerlite in #16209
- fix(anthropic-adapter): properly translate Anthropic image format to OpenAI by @Nivg in #16202
- Feat: Enable automated prompt caching message format for Claude on Databricks by @anthonyivn2 in #16200
- Add gemini live audio model cost in model map by @Sameerlite in #16183
- Fix translation problem with Gemini parallel tool calls by @pablobgar in #16194
- feat: Generalize tiered pricing in generic cost calculator by @AlanPonnachan in #16150
- [MCP] configure static mcp header by @uc4w6c in #16179
- add: benchmark comparison with other gateways by @AlexsanderHamir in #16145
- Revert "add: benchmark comparison with other gateways" by @krrishdholakia in #16247
- [Feature] UI - Litellm test key audio by @yuneng-jiang in #16251
- [Feature] UI - Prevent trailing slash in sso proxy base url input by @yuneng-jiang in #16244
- [Feature] UI - Tag Usage Top Model Table View and Label Fix by @yuneng-jiang in #16249
- fix(proxy): handle None values in daily spend sort key by @AlexsanderHamir in #16245
- fix: Fix OpenAI Responses API streaming tests usage field names and cost calculation by @Chesars in #16236
- [Feature] UI - Initial changes for supporting prompts to multiple models by @yuneng-jiang in #16223
- Add tags and descriptions support to aws secrets manager by @deepanshululla in #16224
- [Feat] Add Bedrock Agentcore as a provider on LiteLLM Python SDK and LiteLLM AI Gateway by @ishaan-jaff in #16252
- [Feat] /search API - add firecrawl search API support by @ishaan-jaff in #16257
- [Feat] add serxng search API provider by @ishaan-jaff in #16259
- feat: Support
reasoning_effortfor watsonx chat models by @AlanPonnachan in #16261 - add: shared_session support to responses API by @AlexsanderHamir in #16260
- fix: typo of the word orginal by @bowenliang123 in #16255
- fix(openai): Remove automatic summary from reasoning_effort transformation by @Chesars in #16210
- fix(redis): handle float redis_version from AWS ElastiCache Valkey by @AlexsanderHamir in #16207
- Fix langfuse input tokens logic for cached tokens by @fabriciojoc in #16203
- Add gpt-image-1-mini pricing for azure by @Sameerlite in #16182
- Fix: Send Gemini API key via x-goog-api-key header with custom api_base by @Sameerlite in #16085
- Litellm noma guardrail support images by @dean-zavad in #16199
- [Feat] Add CyberArk Secrets Manager Integration by @ishaan-jaff in #16278
- [Feat] Cyber Ark - Add Key Rotations support by @ishaan-jaff in #16289
- Add allowing Key based prefix to s3 path by @deepanshululla in #16237
- Add: benchmark comparison with other AI gateways by @AlexsanderHamir in #16248
- [Feat] Email Notifications - Ensure Users get Key Rotated Email by @ishaan-jaff in #16292
- [Fix] UI - Fixed Label for vLLM in Model Create Flow by @yuneng-jiang in #16285
- feat: Add configurable mount name and path prefix for HashiCorp Vault by @AlexsanderHamir in #16253
- Stabilize main branch by @Sameerlite in #16243
- OpikLogger: fix the bug with not incorrect attachment to existing trace & refactor by @alexkuzmik in #15529
- [Feat] Add Custom Secret Manager - Allow users to define and write a custom secret manager by @ishaan-jaff in #16297
- [Feature] UI - Include Model Access Group Models on Team Models Table by @yuneng-jiang in #16298
- Gracefully handle connection closed errors during streaming by @AlexsanderHamir in #16294
- [Fix] Litellm non root docker Model Hub Table fix by @yuneng-jiang in #16282
- [Fix] /model_group/info Returning Entire Model List for SSO Users by @yuneng-jiang in #16296
- Update MCP version from 1.10.1 to 1.20.0 by @Granine in #16281
- feat: add parallel execution handling in during_call_hook by @Oodapow in #16279
- fix: Resolve MyPy type checking errors and CI linting by @Chesars in #16277
- [LiteLLM-16250] Proxy to Bedrock will add name to file content, breaks when cache_control in use. by @jgoodyear in #16275
- Add azure gpt-5-pro in model map by @Sameerlite in #16235
- Revert "Update MCP version from 1.10.1 to 1.20.0" by @krrishdholakia in #16300
- Remove Grok 4 Models Reasoning Effort Parameter by @Sameerlite in #16265
- Implemented native XAI Responses API support in LiteLLM by @Sameerlite in #16310
- [Fix] UI - Add Azure AD Token field and Azure API Key optional by @yuneng-jiang in #16331
- Update container documentation to be similar to others by @Sameerlite in #16327
- Fix: Filter AWS authentication params from Bedrock InvokeModel request body by @Sameerlite in #16315
- [Feat] OTEL - Log Cost Breakdown on OTEL Logger by @ishaan-jaff in #16334
- fix: Pass extra_body parameters to provider in Responses API requests by @Chesars in #16320
- [Feat] Add LiteLLM Gateway built in guardrail by @ishaan-jaff in #16338
- [Feat] UI - Allow configuring LiteLLM Custom Guardrail by @ishaan-jaff in #16339
- Revert "Update perplexity cost tracking" by @ishaan-jaff in #16345
- Update Fireworks audio endpoints to new
api.fireworks.aidomains by @Sameerlite in #16346 - [Fix] UI - SSO Proxy Base URL input validation and remove normalizing / by @yuneng-jiang in #16332
- fix: Use valid CallTypes enum value in embeddings endpoint by @Chesars in #16328
- fix: pass aws_region_name in litellm_params by @Qanpi in #16321
- fux verify ssl on sts boto3 by @ShimonMimoun in #16313
- Add
retry-afterheader support for errors502,503,504by @andriykislitsyn in #16288 - feat(guardrails): panw prisma airs guardrail deduplication and enhanced session tracking by @jroberts2600 in #16273
- Add
vertex_credentialssupport tolitellm.rerank()for Vertex AI by @Sameerlite in #16266 - fix global.anthropic.claude-haiku-4-5-20251001-v1:0 supports_reasoning by @reckless-huang in #16263
- [Feature] UI - Litellm usage date picker by @yuneng-jiang in #16264
- Fix: Azure GPT-5 incorrectly routed to O-series config (temperature parameter unsupported) by @xingyaoww in #16246
- [Fix] UI - Remove encoding_format in request for embedding models by @yuneng-jiang in #16367
- Fix databricks streaming by @chenmoneygithub in #16368
- [Fix] UI - Revert Changes for Test Key Multiple Model Select by @yuneng-jiang in #16372
- Add Vertex MiniMAX m2 by @emerzon in #16373
- fix(vertex_ai): Correctly map 429 Resource Exhausted to RateLimitError by @AlanPonnachan in #16363
- Fix MyPy errors for aembedding call_type - CI pass by @Chesars in #16360
- [Security Fix] Remove tornado test files (including test.key), fixes Python 3.13 security issues by @ishaan-jaff in #16342
- [Feat] Secret Manager - Hashicorp, add auth via approle by @ishaan-jaff in #16374
- [Feature] UI - Surface SSO Create errors on create flow by @yuneng-jiang in #16369
- [Feat] Add DD Agent Host support for
datadogcallback by @ishaan-jaff in #16379 - Fix Azure DALL-E-3 health check content policy violation by using safe default prompt by @Sameerlite in #16329
- [Feat] LiteLLM Guardrail - UI Fix, ensure you can see UI Friendly name for PII Patterns by @ishaan-jaff in #16382
- [Feat] Guardrails - LiteLLM Content Filter, Allow Viewing/Editing Content Filter Settings by @ishaan-jaff in #16383
- [UI] Guardrails - allow updating guardrails through UI. Ensure litellm_params actually get updated in memory by @ishaan-jaff in #16384
- Revert "[Fix] UI - Revert Changes for Test Key Multiple Model Select … by @ishaan-jaff in #16386
- docs: fix image generation response format - use 'images' array instead of 'image' object by @Chesars in #16378
- Propagate cache creation/read token costs for model info to fix Anthropic long context cost calculations by @wwwillchen in #16376
- Add cohere embed-v4.0 model support by @Sameerlite in #16358
- Fix cache_read_input_token_cost for gemini-2.5-flash by @stembe-digitalex in #16354
- Fix HostedVLLMRerankConfig will not be used by @jfcherng in #16352
- Fix Gemini minimal reasoning env overrides disabling thoughts by @Sameerlite in #16347
- Feat/persist mcp credentials in db by @uc4w6c in #16308
- feat: Add support for Anthropic Memory Tool by @AlanPonnachan in #16115
- [Fix] UI - Test Key Page Revert Model To Single Select by @yuneng-jiang in #16390
- [Infra] Building UI for sanity testing by @yuneng-jiang in #16399
- [Feature] UI - Cache Settings Redis Add Semantic Cache Settings by @yuneng-jiang in #16398
- fix(pass_through_endpoints.py): improve clearing logic - only remove unvisited endpoints by @krrishdholakia in #16400
- Added xai responses support by @Sameerlite in #16391
- docs: Add documentation for Anthropic Memory Tool by @AlanPonnachan in #16388
- (feat) audio transcriptions cost tracking (for azure/other non-openai models) + (fix) image generations - accurate cost tracking for dalle3/gpt-image-1 - uses the correct max image quality by @krrishdholakia in #16076
- Revert "(feat) audio transcriptions cost tracking (for azure/other non-openai models) + (fix) image generations - accurate cost tracking for dalle3/gpt-image-1 - uses the correct max image quality" by @krrishdholakia in #16402
New Contributors
- @steve-gore-snapdocs made their first contribution in #16149
- @timbmg made their first contribution in #16120
- @Nivg made their first contribution in #16202
- @pablobgar made their first contribution in #16194
- @AlanPonnachan made their first contribution in #16150
- @Chesars made their first contribution in #16236
- @bowenliang123 made their first contribution in #16255
- @dean-zavad made their first contribution in #16199
- @alexkuzmik made their first contribution in #15529
- @Granine made their first contribution in #16281
- @Oodapow made their first contribution in #16279
- @jgoodyear made their first contribution in #16275
- @Qanpi made their first contribution in #16321
- @ShimonMimoun made their first contribution in #16313
- @andriykislitsyn made their first contribution in #16288
- @reckless-huang made their first contribution in #16263
- @chenmoneygithub made their first contribution in #16368
- @stembe-digitalex made their first contribution in #16354
- @jfcherng made their first contribution in #16352
Full Changelog: v1.79.1-nightly...v1.79.2-nightly