scriptoriumadmin/litellm
1
0
Fork 0
mirror of https://github.com/BerriAI/litellm.git synced 2026-09-29 01:42:19 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 79
41449 commits 2937 branches 1424 tags 3.8 GiB
Commit graph

41449 commits

This branch
This branch
All branches
Author SHA1 Message Date
Harshit Jain
424a96580b
Merge pull request #22627 from BerriAI/litellm_presidio_antropic_edge_case 
fix: presidio edge case with antropic handle on pii token leak
 2026-03-10 21:16:56 +05:30
Harshit Jain
e1c2c36f67
Merge pull request #23259 from BerriAI/litellm_claude_missing_reqs 
fix: claude code req traces on langfuse
 2026-03-10 21:15:36 +05:30
Harshit Jain
fdb0a46242
Merge pull request #23209 from BerriAI/litellm_silent_metrics 
feat: record silent metrics
 2026-03-10 21:14:09 +05:30
Harshit Jain
a7e9bca103
Merge pull request #23214 from BerriAI/litellm_deployment_helm_chart 
feat: add strategy to deployment for helmchart
 2026-03-10 21:13:40 +05:30
Chesars
8c2c379cb5 fix(openai): add missing gpt-5.3 model entry 
Fixes #23267 — plain `gpt-5.3` was missing from the model pricing
JSON, causing tool_choice (and other capability flags) to default
to unsupported. Copied fields from gpt-5.3-chat-latest.
 2026-03-10 12:35:23 -03:00
Carlo Alberto Ferraris
323b473835
fix: add missing indexes for top CPU-consuming queries (#23147) 
* fix: add missing indexes for top CPU-consuming queries

Add indexes to eliminate full table scans on two of the top 5 queries
by CPU usage:

1. LiteLLM_VerificationToken(key_alias) — for ORDER BY key_alias ASC
   queries when listing verification tokens
2. LiteLLM_SpendLogs(user, startTime) — for WHERE user = $1 AND
   startTime BETWEEN $2 AND $3 GROUP BY queries on the spend logs

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* fix: use CREATE INDEX CONCURRENTLY to avoid table locks

Both indexes are now created with CONCURRENTLY and IF NOT EXISTS
to avoid blocking writes on large production tables.
Uses -- SkipTransactionBlock for Prisma migrate compatibility.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
 2026-03-10 21:00:22 +05:30
Chesars
d232d0de6c docs(openai): document gpt-5.4 reasoning_effort + tools limitation 
Add tip boxes explaining that gpt-5.4 does not support reasoning_effort
with function tools in /v1/chat/completions, and that the responses
bridge (openai/responses/gpt-5.4) should be used instead.
 2026-03-10 12:04:55 -03:00
Chesars
08d81f5d7c fix(vertex): shallow copy parameters before mutating in _build_vertex_schema_for_gemini_2 
Avoids silently removing $defs from the caller's dict, which could
affect logging, caching, or retry logic referencing the same object.
 2026-03-10 11:29:59 -03:00
Aarish Alam
2b093aa796
Merge pull request #23196 from CAFxX/docs/claude-md-db-performance-guidelines 
docs: add DB query performance guidelines to CLAUDE.md
 2026-03-10 19:51:51 +05:30
Chesars
a9c3095cc5 fix(vertex): skip harmful schema transforms for Gemini 2.0+ tool parameters 
Gemini 2.0+ natively accepts JSON Schema in tool parameters, including
bare {} (TYPE_UNSPECIFIED), anyOf with null, and lowercase types. The
existing _build_vertex_schema pipeline was coercing {} to {"type": "object"},
breaking JsonValue/Any field semantics (issue #22391).

Add _build_vertex_schema_for_gemini_2() that only resolves $ref (which
Gemini doesn't support in tools) and filters unsupported fields. Use it
for Gemini 2.0+ models, keeping the full transform for Gemini 1.5.
 2026-03-10 11:15:27 -03:00
Chesars
a6cb510703 merge: resolve conflicts between main and litellm_oss_staging_03_04_2026 
Resolved 14 file conflicts:
- image_edits.md: combined OpenRouter + Black Forest Labs providers
- utils.py: kept staging's message-level cache_control check
- networking.tsx: kept export on 4 tool interfaces
- tool_management_endpoints.py: kept ToolOutputPolicy import
- Accepted main's version for: schema.prisma, a2a_protocol, mcp_server,
  _types.py, auth_checks.py, db_spend_update_writer, endpoints.py,
  spend_tracking_utils, a2a_endpoints, model_prices backup
 2026-03-10 10:45:04 -03:00
Chesars
8fac04208d docs(responses): add tool_search bridge examples for chat completions 
Add examples showing tool_search with namespaces via the chat
completions bridge (openai/responses/ prefix) for both SDK and proxy.
 2026-03-10 10:27:40 -03:00
Harshit Jain
17804edc78
Update litellm/utils.py 
Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com>
 2026-03-10 18:44:57 +05:30
Harshit Jain
81bd62e8b0
Update litellm/utils.py 
Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com>
 2026-03-10 18:39:40 +05:30
Harshit28j
558523fd75 fix: req changes by greptile 2026-03-10 18:33:17 +05:30
Harshit28j
2f5a553a7d test: assert setup_database called with correct args 
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
 2026-03-10 18:31:04 +05:30
Harshit28j
8c3d6db482 fix: claude code req traces on langfuse 2026-03-10 18:21:42 +05:30
Chesars
bec12db635 docs(responses): add tool_search & namespaces section for gpt-5.4 
Add documentation for OpenAI's tool_search feature (Responses API)
with SDK and Proxy examples showing namespace-based deferred tool
loading. Closes #23206.
 2026-03-10 09:50:08 -03:00
Sameer Kankute
a788b21092
Merge pull request #23243 from BerriAI/litellm_bedrock-completion-tokens-details 
fix(bedrock): populate completion_tokens_details in Responses API
 2026-03-10 18:19:28 +05:30
Sameer Kankute
cd80213f1c
Merge pull request #23151 from BerriAI/litellm_preserve-reasoning-summary-for-responses-api 
fix(openai): preserve reasoning_effort summary field for Responses API
 2026-03-10 18:18:46 +05:30
Sameer Kankute
b3c4326845
Merge pull request #23131 from BerriAI/litellm_vertex_ai_extra_body_cache_fix 
fix(vertex_ai): strip LiteLLM-internal keys from extra_body before merging to Gemini request
 2026-03-10 18:17:29 +05:30
Harshit28j
aae2deb839 fix: remove redundant import and add test for startup failure 
- Remove redundant `import sys` (already imported at module level)
- Add test_startup_fails_when_db_setup_fails verifying sys.exit(1)
  when PrismaManager.setup_database returns False

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
 2026-03-10 18:15:11 +05:30
Harshit28j
4714128da5 fix: fail proxy startup if prisma migrate fails 
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
 2026-03-10 18:05:13 +05:30
Sameer Kankute
9bc7357e7c
Merge pull request #23222 from BerriAI/litellm_oss_staging_02_18_2026 
Litellm oss staging 02 18 2026
 2026-03-10 17:46:06 +05:30
Sameer Kankute
b211b67cea
Merge pull request #23164 from BerriAI/litellm_oss_staging_03_09_2026 
oss staging 03/09/2026
 2026-03-10 17:45:17 +05:30
Sameer Kankute
d0d1291c15
Merge pull request #22628 from BerriAI/litellm_oss_staging_03_02_2026 
Litellm oss staging 03 02 2026
 2026-03-10 17:40:31 +05:30
Sameer Kankute
9bee76c30d
Merge pull request #23253 from BerriAI/main 
merge main
 2026-03-10 17:25:01 +05:30
Sameer Kankute
72fc2f77cf
Merge pull request #23252 from BerriAI/main 
merge main
 2026-03-10 17:24:19 +05:30
Harshit Jain
9f15a9915e
Merge branch 'main' into litellm_langfuse_key_leakage 2026-03-10 17:21:15 +05:30
Sameer Kankute
3f17a63b81
Merge branch 'main' into litellm_oss_staging_03_02_2026 2026-03-10 17:19:37 +05:30
Harshit28j
20bf3aa807 fix: pop sensitive keys from langfuse 2026-03-10 17:18:11 +05:30
Harshit28j
631aefea18 fix: req changes on feedback from greptile 2026-03-10 16:46:18 +05:30
Harshit28j
212059cd11 fix: presidio improvements 2026-03-10 16:33:41 +05:30
Harshit28j
12de8a724f fix: clean approach instead of UUID 2026-03-10 15:51:11 +05:30
Harshit28j
14ecc79760 fix: req changes greptile hallucinates 2026-03-10 15:51:11 +05:30
Harshit28j
a28fbba3b1 fix: req changes to improve score 2026-03-10 15:51:11 +05:30
Harshit28j
1bfd88a33c fix: req changes 2026-03-10 15:51:10 +05:30
Harshit28j
1ba42d1d99 fix: address req changes 2026-03-10 15:51:10 +05:30
Harshit28j
861db111a0 fix: presidio edge case with antropic handle on pii token leak 2026-03-10 15:51:10 +05:30
Sameer Kankute
94b002066e
Merge pull request #23244 from BerriAI/litellm_skills_api_folder_name_match 
fix(skills): match zip folder name to skill name in SKILL.md for Anthropic API
 2026-03-10 14:11:49 +05:30
Sameer Kankute
88c4079b16 fix(skills): match zip folder name to skill name in SKILL.md for Anthropic API 
Made-with: Cursor
 2026-03-10 14:11:11 +05:30
Sameer Kankute
0122abfa2b fix(bedrock): populate completion_tokens_details in converse _transform_usage 
Co-Authored-By: Claude Sonnet 4.6 (1M context) <noreply@anthropic.com>
 2026-03-10 13:17:28 +05:30
Harshit28j
afb117f5d3 fix: req changes from greptile 2026-03-10 13:03:29 +05:30
Harshit28j
b731e432f8 feat: record silent metrics 2026-03-10 12:54:32 +05:30
Sameer Kankute
3e1f343d1a
Merge pull request #23242 from BerriAI/litellm_fix_redaction_and_charity_engine_provider 
fix(tests): update cache hit redaction assertion to expect choices fo…
 2026-03-10 12:27:59 +05:30
yuneng-jiang
d39a3a90b4
Merge pull request #23241 from BerriAI/litellm_/elastic-gagarin 
[Feature] UI - Virtual Keys: Add refetch button, keep stale data during refetch
 2026-03-09 23:52:11 -07:00
yuneng-jiang
129ed1014b [Fix] UI - Virtual Keys: Reset refetch button on error so user can retry 
When a fetch fails, the button now exits the loading state instead of
staying stuck on "Fetching" indefinitely.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
 2026-03-09 23:46:45 -07:00
Sameer Kankute
f33f446476
Merge pull request #23240 from BerriAI/litellm_bedrock-nova-output-config-spend-tracking 
fix(bedrock): strip output_config from Converse requests; fix spend tracking redaction test
 2026-03-10 12:15:06 +05:30
Sameer Kankute
30fde1de7f fix(tests): update cache hit redaction assertion to expect choices format 
Made-with: Cursor
 2026-03-10 12:14:24 +05:30
yuneng-jiang
9ee489863d [Feature] UI - Virtual Keys: Add refetch button and keep stale data during refetch 
Show a Fetch/Fetching button next to "Showing X of Y results" that acts as
both a manual refetch trigger and a loading indicator. The "Loading keys..."
message now only appears on initial load; subsequent refetches keep the table
visible with stale data (via React Query's keepPreviousData).

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
 2026-03-09 23:30:55 -07:00