* fix: using litellm with claude code bedrock
* fix: usage for bedrock with /messages
* fix: bedrock_sse_wrapper
* tests: test for test_chunk_parser_usage_transformation
* test fix
* fix(huggingface): use get() instead of pop() for input_type parameter
Fixes embedding generation for HuggingFace models where input_type override
is required (e.g. BAAI/bge-m3). The pop() method was mutating optional_params
and removing input_type before downstream functions could access it.
* Add unit tests to catch regression
* Move tests around
* fix(convert_dict_to_response.py): handle None values in usage field for gpt-image-1
* test: add tests for handling None and partial values in usage fields for gpt-image-1 responses
- Renamed SSOSettingsResponse to inherit from a new base class SettingsResponse for better structure.
- Introduced InternalUserSettingsResponse and DefaultTeamSettingsResponse models for internal user and default team settings.
- Updated endpoint responses to use field_schema instead of schema for consistency.
- Enhanced test cases to validate the new response structure and ensure proper functionality of SSO settings.
* fix(anthropic/): fix regression when passing file url's to the 'file_id' parameter
add test and ensure anthropic file url's are correctly sent as 'document' blocks
* fix(vertex_and_google_ai_studio.py): Use same usage calculation function as non-streaming
Closes https://github.com/BerriAI/litellm/issues/10667
* test(test_vertex_and_google_ai_studio_gemini.py): update test
* fix(ui_sso.py): update user as proxy admin in db table, when checking for proxy_admin_id
Fixes issue where existing internal user, unable to make calls when set as proxy admin id
* fix(utils.py): fix custom base path
* Fixes issue with team_endpoints on member budget update
* refactored location of budget membership fix
* added test for _upsert_budget_membership func
* fix(proxy_server.py): working swagger on custom base
removes the swagger monkey patch - this seems to render the swagger on custom base paths
* fix(ui/): working custom auth uptil login success event
* fix(ui/): working custom server root path for login
* fix(proxy_server.py): create typed dict for ui returned token
allows better documentation of expected params
* refactor(proxy_server.py): refactor all ui login endpoints to use same returned ui token object
* feat(ui_sso.py): add server root path to ui token
* feat(ui_sso.py): allows ui to call correct endpoint
* fix(networking.tsx): update proxy base url with custom root path
* fix(networking.tsx): handle updating proxy base url for non-local instances
* refactor: remove uneccessary references to proxybaseurl in ui code - reduce potential for errors
* fix: fix linting error
* fix(onboarding_link.tsx): fix onboarding link when custom server path is set
* feat(ui_discovery_endpoints.py): add new public .well-known/ route for litellm ui config
returns the server root path and proxy base url for constructing api calls
* feat(_types.py): add litellm well known config as public route
allows ui to query it
* fix(/_types.py): add .well-known config to as public route
* fix(page.tsx): create pattern for loading in ui config before making network requests
ensures requests are formatted correctly
* fix(page.tsx): call credential endpoint once ui config is loaded
* fix(page.tsx): route correctly to litellm dashboard from new user login
* fix(page.tsx): remove hardcoded `/litellm` for /sso/key/generate request
* fix(proxy_server.py): re-add moderations endpoint
* fix(proxy_server.py): mount __next__ at / and /litellm
allows it to work when proxy is mounted on root
* docs(contributing.md): remove /ui on ui doc - it will now run on root
* docs(custom_root_ui.md): add docs on custom root path
- Changed hardcoded model "whisper-1" to dynamic model extraction in AzureAudioTranscription and OpenAIAudioTranscription classes.
- Added tests to ensure correct model mapping for various transcription models, including GPT-4o and Azure whisper-1.
* feat: Add presidio_language parameter for PII analysis configuration via yaml config files
- Introduced presidio_language parameter in guardrail_initializers and guardrail_hooks to allow language specification for Presidio PII analysis via yaml config files
- Updated PresidioPresidioConfigModelUserInterface to include presidio_language with a default value of "en".
- Added tests to verify language configuration functionality, ensuring correct language usage in analyze requests.
* docs: Enhance PII masking documentation with language support details
- Added information on configuring the default language for PII analysis using the `presidio_language` parameter in both the UI and YAML configuration.
- Included supported language codes (English, Spanish, German) and their precedence in language settings.
- Provided examples for default language configuration and per-request language overrides to clarify usage.
- Updated model_prices_and_context_window.json to include embed-v4.0 with relevant pricing and metadata.
- Added embed-v4.0 to cohere_embedding_models in constants.py.
- Implemented comprehensive tests for Cohere Embed v4.0 in test_cohere.py, covering basic functionality, input types, error handling, and optional parameters.
* Add support for DataRobot as a router in LiteLLM
* Updates to logic
* Changes to make things work better
* Capitalize bearer
* Revert change
* Undo and simplify things
* Add basic testing
* Add some extra handling
* More tests
* Lowercase
* Fix
* Comment
* Add local test_completion for datarobot
* Sync repo to main
* Update get_complete_url to accept deployments
* Migrate to OpenAILike
* Mock datarobot in test
* Migrate everything to OpenAI LIke
* Apply fixes and changes from review
* Update docs
* Update env vars
* Move tests
* fix(base_routing_strategy.py): compress increments to redis - reduces write ops
* fix(base_routing_strategy.py): make get and reset in memory keys atomic
* fix(base_routing_strategy.py): don't reset keys - causes discrepency on subsequent requests to instance
* fix(parallel_request_limiter.py): retrieve values of previous slots from cache
more accurate rate limiting with sliding window
* fix: fix test
* fix: fix linting error
* feat: Add Gemini TTS audio parameter support
- Add is_model_gemini_audio_model() method to detect TTS models
- Include 'audio' parameter in supported params for TTS models
- Map OpenAI audio parameter to Gemini speechConfig format
- Add _extract_audio_response_from_parts() method to transform audio
output to openai format
* updated unit-test to use pcm16
* - created typedict for speechconfig
- simplified gemini tts model detection
- moved gemini_tts test to test_litellm
* simplified is_model_gemini_audio_model more
* feat(parallel_request_limiter_v2.py): add sliding window logic
allows rate limiting to work across minutes
* fix(parallel_request_limiter_v2.py): decrement usage on rate limit error
* fix(base_routing_strategy.py): fix merge from redis - preserve values in in-memory cache during gap b/w push to redis and read from redis
* fix(base_routing_strategy.py): catch the delta change during redis sync
ensures values are kept in sync
* fix(parallel_request_limiter_v2.py): update tpm tracking to use slot key logic
* fix: fix linting error
* test: update testing
* test: update tests
* test: skip on rate limit or internal server errors
* test: use pytest fixture instead
* test: bump mistral model
