* feat(mcp): BYOK (Bring Your Own Key) for OpenAPI MCP servers with OAuth 2.1 flow
Adds per-user credential storage for BYOK MCP servers so external clients
can authenticate via standard OAuth 2.1 PKCE without needing a full identity
provider.
Backend:
- New DB table LiteLLM_MCPUserCredentials (user_id, server_id, credential_b64)
- is_byok, byok_description, byok_api_key_help_url fields on MCPServerTable
- OAuth 2.1 authorization server endpoints (/.well-known/oauth-authorization-server,
/.well-known/oauth-protected-resource, /v1/mcp/oauth/authorize, /v1/mcp/oauth/token)
- 401 challenge with WWW-Authenticate header when BYOK server has no credential
- CRUD endpoints: POST/DELETE /v1/mcp/server/{id}/user-credential
- has_user_credential annotated on GET /v1/mcp/server response
UI:
- ByokCredentialModal: 2-step Connect flow (access description + API key entry)
- BYOK toggle + description fields on admin MCP server create form
- Connect/Connected state in MCP server table
- BYOK Demo page (/tools/byok-demo) showing full OAuth 2.1 PKCE flow
* feat(mcp/byok): redesign OAuth authorize page to match 2-step Connect mockup
- Step 1: L→S logos, requested access checklist, How it works box, Continue button
- Step 2: API key input, Save toggle, Duration pills (1h/24h/7d/30d/until_revoked), security note
- Matches screenshots: white modal on dark bg, progress dots, dark CTA buttons
- Authorize handler now fetches byok_description and byok_api_key_help_url from server registry
- CLAUDE.md: replace SQL snippet with proper DB migration troubleshooting guidance
* fix: address greptile review feedback (greploop iteration 1)
- XSS: escape all user-supplied values in _build_authorize_html() with html.escape()
- Open redirect: validate redirect_uri scheme and URL-encode code/state in redirect
- N+1 query: batch BYOK credential lookup into single find_many() call
- Critical path DB: add 60s TTL in-memory cache to _check_byok_credential()
- Encrypt BYOK credentials at rest using encrypt_value_helper/decrypt_value_helper
* fix(byok): update OAuth popup with LiteLLM logo, MCP title suffix, remove emojis
* fix(byok-demo): fix token endpoint URL (/v1/mcp/oauth/token not /v1/mcp/token)
* feat(byok): inject stored BYOK credential as mcp_auth_header on tool execution
* feat(byok): use contextvars to inject per-user credential into OpenAPI tool closures; remove byok-demo from LiteLLM UI
OpenAPI tools have auth headers baked into their closures at registration time. BYOK servers have
no static auth token, so per-user credentials were never reaching the HTTP calls.
Fix: add _request_auth_header ContextVar in openapi_to_mcp_generator.py. create_tool_function now
reads this var at call time and overrides the Authorization header if set. execute_mcp_tool resolves
the MCP server and performs BYOK checks before the local-tool dispatch branch, then sets the
ContextVar around _handle_local_mcp_tool so the credential flows into the HTTP request.
Also remove the /tools/byok-demo page from the LiteLLM UI dashboard — the demo lives at
~/Downloads/litellm-byok-demo/index.html (served separately on port 8080).
* fix: address greptile review feedback (greploop iteration 2)
- Cache invalidation: add _invalidate_byok_cred_cache() and call it after
store_user_credential() in both token endpoint and management endpoint
- Unbounded cache: add _BYOK_CRED_CACHE_MAX_SIZE=4096 with clear-on-overflow
- Unbounded auth codes: add _AUTH_CODES_MAX_SIZE=1000 with 503 on overflow
- Double DB query: merge _check_byok_credential + _get_byok_credential into
single _get_byok_credential call; raise 401 inline if None returned
- Sidebar: remove byok-demo entry (page was deleted in prior commit)
- JWT comment: document why byok_session HS256 token can't be used as proxy auth
* fix: address greptile review feedback (greploop iteration 3)
- auth_type: pre-format Authorization header (Bearer/ApiKey/Basic) in server.py
before setting ContextVar so openapi_to_mcp_generator respects server auth_type
- cache invalidation on delete: call _invalidate_byok_cred_cache after
delete_user_credential so stale True entries don't persist for 60s
- ContextVar guard: only set _request_auth_header when mcp_auth_header is set,
avoiding unnecessary ContextVar overhead on non-BYOK tool calls
* fix: address greptile review feedback (greploop iteration 4)
- Unified credential cache: store actual credential value (Optional[str])
instead of just bool so _get_byok_credential also benefits from caching —
eliminates the DB hit on every BYOK tool call within the 60s TTL window
- Extracted _write_byok_cred_cache() helper for consistent cache writes
- Replaced has_user_credential with get_user_credential in _check_byok_credential
so one DB call satisfies both existence check and value retrieval
- Remove false 'encrypted at rest' claim from OAuth HTML and ByokCredentialModal
* Update tests/test_litellm/proxy/_experimental/mcp_server/test_byok_oauth_endpoints.py
Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com>
* Update tests/test_litellm/proxy/_experimental/mcp_server/test_byok_oauth_endpoints.py
Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com>
---------
Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com>
* fix(gemini): ensure image token accumulation in usage metadata
Fixed an issue where image tokens were being overwritten instead of accumulated in Gemini responses. Added support for both camelCase and snake_case token count keys. Fixes#22082.
* test: add regression test for image token accumulation and cleanup files
* fix(gemini): ensure consistent accumulation for responseTokensDetails
* fix(gemini): harden token count parsing and add vertex accumulation test
Parse tokenCount/token_count as int-safe values to satisfy mypy and avoid None/object arithmetic. Add regression test for duplicate modality accumulation in Vertex _calculate_usage.
* fix: complexity_router fails on list-format message content (OpenAI multi-part messages)
When a client sends messages with list-format content
(e.g. [{"type": "text", "text": "..."}] as used by the OpenAI JS SDK
and other clients), the complexity_router's async_pre_routing_hook
skipped those messages because it only handled str content. This caused
user_message to be None, the hook returned None, and the router fell
through to selecting the complexity_router deployment itself
(model="auto_router/complexity_router") which litellm cannot dispatch,
resulting in LiteLLMUnknownProvider.
Fixes:
- Extract text from list-format content parts (type=text) before
classifying
- Return default_model instead of None when no user message can be
extracted, preventing the crash fallthrough
- Loosen PreRoutingHookResponse.messages type from Dict[str, str] to
Dict[str, Any] to accommodate list-format content values
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* fix: update messages type annotation in async_pre_routing_hook to Dict[str, Any]
Consistent with PreRoutingHookResponse.messages type change and the
list-format content support added in the previous commit.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* fix: normalize None content to empty string in complexity_router message parsing
msg.get("content", "") returns None when the key exists with value None
(e.g. assistant messages with tool calls). Use `or ""` to normalize
None to an empty string explicitly.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* fix: strip whitespace from joined list content parts in complexity_router
Prevents leading/trailing spaces when some content parts have empty
text values (e.g. " ".join(["", "hello"]) → " hello").
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
---------
Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
* (sap) ensure tool parameters have type='object' for SAP compatibility
Fix SAP GenAI Hub Orchestration Service rejecting tool calls with error:
"400 - LLM Module: tools.0.custom.input_schema.type: Input should be 'object'"
Root cause: When Claude Code uses tools (like web_search) with the SAP provider
through LiteLLM's Anthropic experimental pass-through adapter, Anthropic's
input_schema format doesn't always include the required type="object" field.
The adapter's translate_anthropic_tools_to_openai() function was directly
copying input_schema to OpenAI's parameters field without ensuring the
type="object" requirement that SAP's API strictly enforces.
Changes:
- Modified translate_anthropic_tools_to_openai() to check if input_schema
is missing the type field and add type="object" if absent
- Preserves existing type field if already present
- Added comprehensive test suite (6 tests) covering:
- Missing type field scenario (now adds type="object")
- Existing type preservation
- Empty input_schema handling
- Multiple tools transformation
- Additional schema properties preservation
- SAP-specific compatibility regression test
Testing:
- All new tests pass (6/6 in test_anthropic_tool_schema_fix.py)
- All existing Anthropic tool tests pass (57/57 tool-related tests)
- SAP tool parameter validation tests pass (9/9 in test_sap_tool_parameters.py)
* (sap) enable native response_format for anthropic models
* (sap) filter strict param from model_params for GPT models only
* (sap) revert Anthropic adapter type='object' fix
The SAP FunctionTool Pydantic validator in litellm/llms/sap/chat/models.py
already ensures type='object' is added to all tool parameters for SAP
API compatibility.
The Anthropic adapter change affected ALL consumers, not just SAP, which
was broader scope than intended for this PR.
- Revert input_schema modification in Anthropic adapter
- Remove Anthropic-specific test file (SAP tests still cover this case)
* (sap) gate markdown stripping to Anthropic models only
SAP GenAI Hub with Anthropic models sometimes returns JSON wrapped in
markdown code blocks. GPT/Gemini/Mistral models don't exhibit this
behavior, so stripping is now gated to avoid accidentally modifying
valid responses that may contain markdown in JSON string values.
The user-specified async client was being overwritten by
`litellm.module_level_aclient` in `streaming_handler.py` when using
async+streaming with Gemini.
This fix adds a `gemini_client` parameter to `make_call()` (matching
the existing pattern in `make_sync_call()`) so the user's custom client
is preserved and not overwritten.
Fixes#17148
- Log provider token counting failures instead of silently swallowing
- Fall back to local tokenizer when provider returns error response
- Map assistant tool_calls to Responses API function_call items
- Concatenate multiple system messages instead of overwriting
- Hide internal error details from proxy API responses
- Narrow exception catch in handler to network/JSON errors only
- Update test to match new fallback behavior
- Add OpenAITokenCounter using POST /v1/responses/input_tokens endpoint
- Add litellm.acount_tokens() public async API that auto-routes to provider APIs
- Add proxy endpoint POST /v1/responses/input_tokens for OpenAI-compatible counting
- Transform chat tools format to Responses API format for correct token counting
- Fall back to local tiktoken when provider API unavailable
Fixes#22302
Kontext models (flux-kontext-pro, flux-kontext-max) support both
text-to-image and image editing. Add them to IMAGE_GENERATION_MODELS
and update supported_endpoints in model prices JSON.
- Create handler.py for image generation and image edit
- Move polling logic from transformation to handlers
- Handlers use _get_httpx_client() / get_async_httpx_client()
- Transformation files now only transform request/response data
- Follows Bedrock pattern for provider-specific handlers
Addresses feedback: transformation files should not make HTTP requests
Add native text-to-image generation for Black Forest Labs Flux models
(flux-pro-1.1, flux-pro-1.1-ultra, flux-dev, flux-pro).
- Polling-based async API with sync and async support
- OpenAI-compatible parameter mapping (size, n, quality)
- Reuses shared HTTP clients via _get_httpx_client()
- 39 unit tests added
Replace direct httpx.get() calls with _get_httpx_client() to reuse
cached HTTP client, following the pattern used by other providers
(RunwayML, Azure AI OCR, Sagemaker, etc.).
Add native integration for Black Forest Labs image editing models
(flux-kontext-pro, flux-kontext-max, flux-pro-1.0-fill, flux-pro-1.0-expand).
Changes:
- Add BlackForestLabsImageEditConfig for BFL API transformation
- Add BLACK_FOREST_LABS to LlmProviders enum
- Add use_multipart_form_data() to BaseImageEditConfig for JSON vs form-data
- Modify image_edit_handler to support JSON request bodies
- Add comprehensive unit tests
Closes#11401
Providers like Cerebras return delta.reasoning in streaming responses
for gpt-oss models, but LiteLLM's Delta class expects reasoning_content.
This causes reasoning content to be silently dropped during streaming.
Fixes#13300
Add global media_resolution support for Gemini 2.x models (2.0, 2.5) when
using OpenAI's detail parameter on images. Previously, the detail parameter
was only working for Gemini 3+ models (per-part) and was silently ignored
for older Gemini models.
- Add _get_highest_media_resolution() and _extract_max_media_resolution_from_messages()
to extract highest detail from all images/files in a request
- Update _transform_request_body() to add mediaResolution to generationConfig
for Gemini 2.x models only (not 1.x which doesn't support it, not 3+ which
uses per-part)
- Add mediaResolution field to GenerationConfig TypedDict
- Support detail extraction from both image_url and file content types
- Add comprehensive unit tests and update documentation
Add MistralAudioTranscriptionConfig for Mistral's /v1/audio/transcriptions
endpoint, enabling litellm.transcription() with mistral/voxtral-mini-latest
and other Voxtral models. Supports multipart form-data with OpenAI-compatible
params (language, temperature, response_format, timestamp_granularities)
plus Mistral-specific params like diarize.
* azure content enhancement...
* rafactored to increase confidence score
* improvements based on additional feedback
* removed unused import
* Force-split any word longer than max length allowed
* preserve whitespace in text splitting
* moving common initialization to base class
* consolidate enforcement into async_make_request as single point, remove redundant caller-side checks, extract shared init/HTTP logic into base, and fix stale log messages
* clean up
* clean up tests
Add cache_read_input_token_cost_per_audio_token, supports_code_execution,
and supports_file_search to the JSON schema used by the model prices
validation test.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The _encrypt_response_id method now receives request_cache=None as a
keyword argument from async_post_call_success_hook. Updated the mock
assertion to expect this parameter.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Tests used call_policy throughout but the actual API model uses
input_policy and output_policy. Updated _make_tool_row helper,
list filter query param, and policy update request/response assertions.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The aresponses_websocket CallType was recently added but not included
in the test exclusion list. It uses WebSocket passthrough (not Azure SDK
client initialization), so it correctly doesn't call
initialize_azure_sdk_client.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Fix translate_thinking_to_reasoning in responses_adapters/transformation.py
to make summary opt-in (was hardcoded to "detailed")
- Update e2e test to mock litellm.responses (new OpenAI routing path)
- Add tests for Responses API adapter summary preservation
- Resolve merge conflict in test file
Remove hardcoded summary="detailed" injection — summary is opt-in per
OpenAI spec and increases costs. Users opt-in per-request via LiteLLM
extension: thinking={"type": "enabled", "budget_tokens": N, "summary": "concise"}.
Also preserve summary in translate_thinking_for_model() which previously
dropped it when converting thinking → reasoning_effort for non-Claude models.
Fixes#20998
Add test_parallel_tool_calls_comprehensive_streaming_integration which
synthesizes the full 10-event Responses API SSE sequence with split
argument deltas and asserts all fix invariants together:
1. output_item.done emits no finish_reason (no premature stream end)
2. Each call_id appears exactly once (no duplicate tool_call chunks)
3. Split argument deltas assemble to correct final JSON
4. Exactly one finish event, at the terminal response.completed chunk
5. Parallel tool calls have distinct indices (output_index 0 and 1)
All 24 unit tests pass.
* add explicit caching to litellm proxy for gemini models via injection
* fix: add missing `supports_function_calling` for deepinfra models
All 55 deepinfra models that had `supports_tool_choice: true` were
missing the `supports_function_calling` flag, causing
`litellm.supports_function_calling()` to incorrectly return False.
Fixes#22619
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* Managed batches - Address PR bot comments from #22464
* feat(togetherai): add support for TogetherAI Qwen3.5-397B-A17B model
* Agent Tracing - support context_id based trace id propogation + nested llm calls (#22626)
* style(ui/): distinguish agent calls from llm calls on ui
* feat: initial grouping working
* feat: set stable contextid for a2a calls - allows for easily passing to downstream llm/mcp calls
* feat(a2a_endpoints.py): fix tracing to avoid recreating logging objects for the same call
allows stable trace id usage
* fix(guardrail_endpoints): handle string ui_type values in _build_field_dict
_build_field_dict unconditionally called .value on ui_type, which crashes
for guardrail configs that use plain strings (e.g. BlockCodeExecutionGuardrailConfigModel
uses "multiselect" and "percentage"). Now checks with hasattr before calling .value.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* fix: propagate trace/session id from headers in MCP server calls
Cherry-picked mcp_server/server.py fixes from 6feb9bab: adds
get_chain_id_from_headers to extract x-litellm-trace-id /
x-litellm-session-id from raw headers, and uses it in call_tool
and list_tools to keep spend logs and tracing consistent with A2A.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
* [Feat] UI - Add Open in New Tab on leftnav Bar (#22731)
* Add minimal dev_config.yaml for proxy development
Co-authored-by: Ishaan Jaff <ishaan-jaff@users.noreply.github.com>
* feat(ui): wrap left nav items in <a> tags for open-in-new-tab support
Nav items are now rendered as <a> elements with proper href attributes,
enabling right-click → 'Open in new tab', Ctrl/Cmd+click, and
middle-click to open any sidebar page in a new browser tab.
Normal clicks continue to use SPA navigation (no full page reload).
Applied to both leftnav.tsx (query-param routing) and Sidebar2.tsx
(Next.js file-based routing).
Co-authored-by: Ishaan Jaff <ishaan-jaff@users.noreply.github.com>
---------
Co-authored-by: Cursor Agent <cursoragent@cursor.com>
Co-authored-by: Ishaan Jaff <ishaan-jaff@users.noreply.github.com>
* [Feat] Add Tool Policies for AI Gateway (#22732)
* fix: fix ui render
* fix: fix minor bugs
* refactor: use prisma functions instead of raw sql (safer)
* fix(add-new-tiles-to-tool-policies): allow developer to see what's available
* feat: ensure tool allowlist runs correctly for tool names + mcp's
* refactor: more ui improvements
* feat: working key tool blocking
* feat(tools): show tool logs
* refactor: backend code improvements
* refactor: improve log viewer for tools
* fix: address PR review feedback for tool access control
- Add missing blocked_tools column to root schema.prisma (schema drift)
- Invalidate ToolPolicyRegistry after policy mutations so changes take effect immediately
- Remove dead code: unused get_effective_policies, get_tool_policies_cached, and helpers
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* fix: race condition in permission resolution and remove duplicate allowlist check
- Use atomic update_many with object_permission_id=None to prevent concurrent
requests from creating orphaned permission rows and losing tool blocks
- Remove duplicate allowed_tools enforcement from guardrail (already enforced
in auth layer via check_tools_allowlist)
- Move inline uuid import to module level
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* update to account for userAgent
* UI - Add ToolDetails
* input/output policy
* LiteLLM_PolicyAttachmentTable
* LiteLLM_PolicyAttachmentTable
* fix: add _enqueue_tool_registry_upsert
* fix: tool mgmt endpoints
* tool mgmt endpoints
* Update tests/test_litellm/proxy/db/test_tool_registry_writer.py
Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com>
* Update tests/test_litellm/proxy/db/test_tool_registry_writer.py
Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com>
* Update tests/test_litellm/proxy/db/test_tool_registry_writer.py
Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com>
* fix: sync root schema.prisma and fix test_tool_registry_writer for input/output policy
- Migrate root schema.prisma LiteLLM_ToolTable from call_policy to
input_policy/output_policy, add missing user_agent and last_used_at columns
(now consistent with litellm/proxy/schema.prisma and litellm-proxy-extras)
- Fix SpendLogToolIndex comment across all three schema files
- Fix all call_policy references in test_tool_registry_writer.py:
swapped update_tool_policy arguments, wrong get_tools_by_names return type
assertions, _mock_tool_row setting call_policy instead of input_policy
Addresses Greptile review feedback on PR #22732.
Made-with: Cursor
---------
Co-authored-by: Krrish Dholakia <krrishdholakia@gmail.com>
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com>
* feat(proxy): add key_alias, key_hash, requested_model DD APM span tags (#22710)
* feat(proxy): add key_alias, key_hash, requested_model tags to DD APM spans
* refactor(proxy): consolidate DD APM tag helpers into DDSpanTagger class
* refactor(proxy): move DDSpanTagger to its own file litellm/proxy/dd_span_tagger.py
---------
Co-authored-by: liweiguang <codingpunk@gmail.com>
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
Co-authored-by: Ephrim Stanley <ephrim.stanley@point72.com>
Co-authored-by: Varad Khonde <varadkhonde@gmail.com>
Co-authored-by: Krish Dholakia <krrishdholakia@gmail.com>
Co-authored-by: Sameer Kankute <sameer@berri.ai>
Co-authored-by: Ishaan Jaff <ishaanjaffer0324@gmail.com>
Co-authored-by: Cursor Agent <cursoragent@cursor.com>
Co-authored-by: Ishaan Jaff <ishaan-jaff@users.noreply.github.com>
Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com>
* feat(proxy): add key_alias, key_hash, requested_model tags to DD APM spans
* refactor(proxy): consolidate DD APM tag helpers into DDSpanTagger class
* refactor(proxy): move DDSpanTagger to its own file litellm/proxy/dd_span_tagger.py
