Commit Graph

21188 Commits

Author SHA1 Message Date
Ishaan Jaff
80fb4ece97 prom emit size of DB TX queues for observability 2025-04-02 18:39:29 -07:00
Ishaan Jaff
7b768ed909 doc fix sso login url 2025-04-02 18:38:33 -07:00
Ishaan Jaff
3256b6af6c track service types on prom services 2025-04-02 18:03:09 -07:00
Ishaan Jaff
05b30e28db clean up service metrics 2025-04-02 17:50:41 -07:00
Ishaan Jaff
73bbd0a446 emit lock acquired and released events 2025-04-02 17:40:25 -07:00
Ishaan Jaff
e09ef4afc7 use service logger for tracking pod lock status 2025-04-02 17:39:48 -07:00
Ishaan Jaff
8405fcb748 test pod lock manager 2025-04-02 15:06:31 -07:00
Ishaan Jaff
8b12a2e5dc fix pod lock manager 2025-04-02 14:52:55 -07:00
Ishaan Jaff
a64631edfb test pod lock manager 2025-04-02 14:39:40 -07:00
Ishaan Jaff
2e939a21b3 refactor pod lock manager to use redis 2025-04-02 14:37:39 -07:00
Krish Dholakia
053b0e741f
Add Google AI Studio /v1/files upload API support (#9645)
* test: fix import for test

* fix: fix bad error string

* docs: cleanup files docs

* fix(files/main.py): cleanup error string

* style: initial commit with a provider/config pattern for files api

google ai studio files api onboarding

* fix: test

* feat(gemini/files/transformation.py): support gemini files api response transformation

* fix(gemini/files/transformation.py): return file id as gemini uri

allows id to be passed in to chat completion request, just like openai

* feat(llm_http_handler.py): support async route for files api on llm_http_handler

* fix: fix linting errors

* fix: fix model info check

* fix: fix ruff errors

* fix: fix linting errors

* Revert "fix: fix linting errors"

This reverts commit 926a5a527ff27a107b39da8f5a26b0ee8e2d9884.

* fix: fix linting errors

* test: fix test

* test: fix tests
2025-04-02 08:56:58 -07:00
Krish Dholakia
453003c378
fix(gemini/): add gemini/ route optional param mapping support (#9677)
Fixes https://github.com/BerriAI/litellm/issues/9654
2025-04-02 08:56:32 -07:00
Krrish Dholakia
6e8d9860a5 ci(config.yml): remove -s from large tests
easier to track test progress
2025-04-02 08:50:47 -07:00
Krrish Dholakia
d32cf141f5 docs: update docs 2025-04-02 07:58:45 -07:00
Pranav Simha
2e35f07e94
Add support for max_completion_tokens to the Cohere chat transformation config (#9701) 2025-04-02 07:50:44 -07:00
Krish Dholakia
6c69ad4c89
fix(model_management_endpoints.py): fix allowing team admins to update team models (#9697)
* fix(model_management_endpoints.py): fix allowing team admins to update their models

* test(test_models.py): add e2e test to for team model flow

ensure team admin can always add / edit / delete team models
2025-04-01 22:28:15 -07:00
Krish Dholakia
3d0313b15b
Litellm user daily activity allow non admin usage (#9695)
* feat(internal_user_endpoints.py): allow non-admin to view their own usage via `/user/daily/activity` route

* fix(leftnav.tsx): allow users to view their own usage via new_usage.tsx

allows internal users to see their usage via new api

Handles 1m+ spend logs scenario

* fix(leftnav.tsx): allow all users to see new usage tab
2025-04-01 22:27:26 -07:00
Krish Dholakia
23051d89dd
fix(streaming_handler.py): fix completion start time tracking (#9688)
* fix(streaming_handler.py): fix completion start time tracking

Fixes https://github.com/BerriAI/litellm/issues/9210

* feat(anthropic/chat/transformation.py): map openai 'reasoning_effort' to anthropic 'thinking' param

Fixes https://github.com/BerriAI/litellm/issues/9022

* feat: map 'reasoning_effort' to 'thinking' param across bedrock + vertex

Closes https://github.com/BerriAI/litellm/issues/9022#issuecomment-2705260808
2025-04-01 22:00:56 -07:00
Tomer Bin
0690f7a3cb
Virtual key based policies in Aim Guardrails (#9499)
* report key alias to aim

* send litellm version to aim

* Update docs

* blacken

* add docs

* Add info part about virtual keys specific guards

* sort guardrails alphabetically

* fix ruff
2025-04-01 21:57:23 -07:00
Ishaan Jaff
ac3399238e
Merge pull request #9690 from BerriAI/litellm_fixes_for_deadlock_round_2
[Reliability Fixes] - Ensure no deadlocks occur when updating `DailyUserSpendTransaction`
2025-04-01 20:59:41 -07:00
Krrish Dholakia
0cfc3e822c build: new ui build 2025-04-01 19:44:23 -07:00
Krrish Dholakia
9f78ddbaec bump: version 1.65.1 → 1.65.2 2025-04-01 19:38:51 -07:00
Krrish Dholakia
09214574b0 build(ui/): update ui build 2025-04-01 19:38:41 -07:00
Krish Dholakia
60863bb134
UI - New Usage Tab fixes (#9696)
* fix(new_usage.tsx): enable smooth scrolling - remove double scroll bars

enables easier viewing of overflowing content

* fix(new_usage.tsx): fix ordering of daily spend

ensure always from earliest to latest date

* feat(internal_user_endpoints.py): return key alias on `/user/daily/activity`

Enables easier consumption on UI

* fix(new_usage.tsx): show key alias on usage tab

* feat(activity_metric.tsx): new activity panel - showing spend per model per day

allows debugging if models are not being tracked

* fix(top_key_view.tsx): use consistent param for selecting key
2025-04-01 19:36:00 -07:00
Ishaan Jaff
feba274a89 test DailySpendUpdateQueue 2025-04-01 18:39:23 -07:00
Ishaan Jaff
4a091a34b0 move test loc 2025-04-01 18:33:33 -07:00
Ishaan Jaff
44bc8974c8 BaseUpdateQueue 2025-04-01 18:31:54 -07:00
Ishaan Jaff
8dc792139e refactor file structure 2025-04-01 18:30:48 -07:00
Ishaan Jaff
290e837515 test_update_logs_with_spend_logs_url 2025-04-01 18:15:01 -07:00
Ishaan Jaff
07fc5a72f1 add debug statement 2025-04-01 17:54:52 -07:00
Ishaan Jaff
827ad38e7f fix spend update queue 2025-04-01 17:49:28 -07:00
Ishaan Jaff
4c14550721 refactor daily spend updates to use new Queue DS 2025-04-01 17:45:19 -07:00
Ishaan Jaff
655ce2e745
Merge pull request #9683 from BerriAI/litellm_fix_service_account_behavior
[Bug fix] - Service accounts - only apply `service_account_settings.enforced_params` on service accounts
2025-04-01 12:04:38 -07:00
Ishaan Jaff
4ddca7a79c
Merge branch 'main' into litellm_fix_service_account_behavior 2025-04-01 12:04:28 -07:00
Ishaan Jaff
61b609f320
Merge pull request #9673 from BerriAI/litellm_qa_deadlock_fixes
[Reliability] - Ensure new Redis + DB architecture tracks spend accurately
2025-04-01 12:04:03 -07:00
Ishaan Jaff
5965680176 fix dev release.txt 2025-04-01 12:02:51 -07:00
Krrish Dholakia
40a792472b build(enterprise.md): add why enterprise to docs 2025-04-01 11:27:03 -07:00
Ishaan Jaff
c2c5dbf24f test_get_enforced_params 2025-04-01 08:41:53 -07:00
Ishaan Jaff
f805e15f7b test_get_enforced_params_for_service_account_settings 2025-04-01 08:39:41 -07:00
Ishaan Jaff
01bd418e0b remove deprecated logic 2025-04-01 07:55:28 -07:00
Ishaan Jaff
e5f6529c42 test_get_enforced_params_for_service_account_settings 2025-04-01 07:46:38 -07:00
Ishaan Jaff
13aa7f75f6 test_enforced_params_check 2025-04-01 07:40:31 -07:00
Ishaan Jaff
f8ce30763e fix check_if_token_is_service_account 2025-04-01 07:36:35 -07:00
Ishaan Jaff
55763ae276 test_end_user_transactions_reset 2025-04-01 07:13:25 -07:00
Ishaan Jaff
7a2442d6c0 test_batch_update_spend 2025-04-01 07:12:29 -07:00
Krrish Dholakia
b0fa934fe3 docs(anthropic.md): update docs with file message usage 2025-03-31 22:58:51 -07:00
Krish Dholakia
62ad84fb64
UI (new_usage.tsx): Report 'total_tokens' + report success/failure calls (#9675)
* feat(internal_user_endpoints.py): return 'total_tokens' in `/user/daily/analytics`

* test(test_internal_user_endpoints.py): add unit test to assert spend metrics and dailyspend metadata always report the same fields

* build(schema.prisma): record success + failure calls to daily user table

allows understanding why model requests might exceed provider requests (e.g. user hit rate limit error)

* fix(internal_user_endpoints.py): report success / failure requests in API

* fix(proxy/utils.py): default to success

status can be missing or none at times for successful requests

* feat(new_usage.tsx): show success/failure calls on UI

* style(new_usage.tsx): ui cleanup

* fix: fix linting error

* fix: fix linting error

* feat(litellm-proxy-extras/): add new migration files
2025-03-31 22:48:43 -07:00
Krish Dholakia
f2a7edaddc
fix(proxy_server.py): Fix "Circular reference detected" error when max_parallel_requests = 0 (#9671)
* fix(proxy_server.py): remove non-functional parent backoff/retry on /chat/completion

Causes circular reference error

* fix(http_parsing_utils.py): safely return parsed body - don't allow mutation of cached request body by client functions

Root cause fix for circular reference error

* Revert "fix: Anthropic prompt caching on GCP Vertex AI (#9605)" (#9670)

This reverts commit a8673246dc.

* add type hints for AnthropicMessagesResponse

* define types for response form AnthropicMessagesResponse

* fix response typing

* allow using litellm.messages.acreate and litellm.messages.create

* fix anthropic_messages implementation

* add clear type hints to litellm.messages.create functions

* fix anthropic_messages

* working anthropic API tests

* fixes - anthropic messages interface

* use new anthropic interface

* fix code quality check

* docs anthropic messages endpoint

* add namespace_packages = True to mypy

* fix mypy lint errors

* docs anthropic messages interface

* test: fix unit test

* test(test_http_parsing_utils.py): update tests

---------

Co-authored-by: Ishaan Jaff <ishaanjaffer0324@gmail.com>
2025-03-31 22:06:02 -07:00
Ishaan Jaff
d0a7e44a6e fix linting 2025-03-31 21:33:05 -07:00
superpoussin22
136f1d60b3
Update model_prices_and_context_window.json add gemini-2.5-pro-exp-03-25 (#9650)
* Update model_prices_and_context_window.json

add gemini-2.5-pro-exp-03-25

* Update model_prices_and_context_window.json

correct tokens limit
2025-03-31 21:31:30 -07:00