Commit graph

12481 commits

Author SHA1 Message Date
Ishaan Jaff
a4cf453ad1 fix - end user rate limiting tests 2024-05-22 16:46:19 -07:00
Krrish Dholakia
9f518030b9 build(schema.prisma): add completion_start_time to spend logs
enables time to first token tracking for models
2024-05-22 16:43:08 -07:00
Rajan Paneru
cc1fcef6e1 reverted with the main to keep them same, and subsequent pr will make these changes 2024-05-23 09:09:04 +09:30
Ishaan Jaff
4175d00a24 fix - test end user rate limits with master key 2024-05-22 16:23:15 -07:00
Ishaan Jaff
a848a676af docs - end user rate limiting 2024-05-22 16:20:25 -07:00
Krrish Dholakia
7f0d85f12a feat(proxy_server.py): enable admin to create/delete budgets for end-users 2024-05-22 16:20:06 -07:00
Ishaan Jaff
42078ac285 fix - run tpm / rpm checks on proxy admin keys too 2024-05-22 16:15:09 -07:00
Ishaan Jaff
e6b406d739 feat - enforce end user tpm / rpm limits 2024-05-22 15:45:30 -07:00
Ishaan Jaff
bef10f4b01 test - end user tpm / rpm limiting 2024-05-22 15:42:41 -07:00
Ishaan Jaff
106910cecf feat - add end user rate limiting 2024-05-22 14:01:57 -07:00
Ishaan Jaff
eac7e70dca feat - include litellm_budget table when getting end_user 2024-05-22 13:30:08 -07:00
Krrish Dholakia
9609df16d3 feat(proxy_server.py): new 'add budget' endpoint
create a budget object (max budget, tpm /rpm limits, etc.) and assign that to a user/team/etc.
2024-05-22 13:16:37 -07:00
Ishaan Jaff
778b41cea5 fix - update migration script to ensure api_key in script 2024-05-22 12:22:23 -07:00
Ishaan Jaff
0623eb0274
Merge pull request #3783 from BerriAI/litellm_fix_dont_use_gunicorn_helm
[Fix] - don't use `gunicorn` on litellm helm
2024-05-22 07:50:47 -07:00
Ishaan Jaff
815dcd9187 fix - don't use gunicorn on litellm helm 2024-05-22 07:48:34 -07:00
Ishaan Jaff
4e5be51373
Merge pull request #3775 from BerriAI/litellm_add_failure_Callbacks_from_db
[Feat] add failure callbacks from DB to proxy
2024-05-22 07:36:04 -07:00
frob
9d4b7c17a6
Merge branch 'BerriAI:main' into ollama-image-handling 2024-05-22 14:08:39 +02:00
vietpham1911
f1ae9f0acd Added the requested features of Merlinvt ("1.If the model is not multimodal, there is no info needed for input_cost_per_image. Right now it's 0.0 for a lot of models.
2. Could you add the field max_output_tokens (is it max_completion_tokens in OpenRouter?).")
2024-05-22 16:10:39 +09:30
Ishaan Jaff
b81fcf2482 feat -add failure callbacks from DB to proxy 2024-05-21 22:00:36 -07:00
Krish Dholakia
5d7d638494
Merge pull request #3770 from BerriAI/litellm_filter_invalid_params
feat(router.py): filter out deployments which don't support request params w/ 'pre_call_checks=True'
2024-05-21 20:42:29 -07:00
Krish Dholakia
beb61702c3
Merge branch 'main' into litellm_filter_invalid_params 2024-05-21 20:42:21 -07:00
Rajan Paneru
4dfbe0be07 ran the script after resolving conflict 2024-05-22 12:15:47 +09:30
Rajan Paneru
2e7c1cb736 resolve-conclit: pulled the changes from main 2024-05-22 12:14:10 +09:30
Rajan Paneru
874847099a moved script to the workflow folder 2024-05-22 12:13:06 +09:30
Rajan Paneru
ac044c4bc6 created github worflow to update the file and create a pull-request 2024-05-22 12:13:06 +09:30
Rajan Paneru
192feba2aa reverted the changes made in circleci 2024-05-22 12:13:03 +09:30
Rajan Paneru
41d76b477b Daily JSON update: 2024-05-19 2024-05-22 12:11:43 +09:30
Rajan Paneru
c0ceac1dba update price & context window file daily, commit and push to the main 2024-05-22 12:11:11 +09:30
Rajan Paneru
51fa42e13a resolve conflict 2024-05-22 12:09:30 +09:30
vietpham1911
1b16652629 Implemented feature to add input_cost_per_image 2024-05-22 12:08:35 +09:30
vietpham1911
4a7fd2ea3e Wrote the code more clean and more efficient. Furthermore, the code adds the input_cost_per_image and supports_vision to the models if the two values exist. 2024-05-22 12:08:35 +09:30
Rajan Paneru
ea4b766a94 updated the config to run the script 2024-05-22 12:08:35 +09:30
vietpham1911
514477b1e7 Bug fix: The prior code removed values like "supports_function_calling" or "supports_vision". This is fixed now.
Nothing is being removed. Only the pricings and the maximum tokens are being edited and new models in openrouter are being added.
2024-05-22 12:08:35 +09:30
vietpham1911
771f06bd17 Add the models of openrouter that are not in the json file with the pricings of the models yet 2024-05-22 12:08:35 +09:30
vietpham1911
5149a1de86 Added script for auto updating the pricing with the pricings of openrouter.ai 2024-05-22 12:08:35 +09:30
Ishaan Jaff
908195610d bump: version 1.37.19 → 1.37.20 2024-05-21 19:11:27 -07:00
Krish Dholakia
febd57dc81
Merge pull request #3768 from BerriAI/litellm_lowest_latency_ttft_routing
feat(lowest_latency.py): route by time to first token, for streaming requests (if available)
2024-05-21 19:11:02 -07:00
Krrish Dholakia
0001b3208d test(test_alangfuse.py): skip langfuse test 2024-05-21 19:00:12 -07:00
Ishaan Jaff
e0cf055be6
Merge pull request #3772 from BerriAI/litellm_create_keys_with_permissions
[Feat] Proxy - Create Keys that can only access `/spend` routes on Admin UI
2024-05-21 18:53:49 -07:00
Ishaan Jaff
b513569a95 docs - invite team members to access /spend routes 2024-05-21 18:53:26 -07:00
Krrish Dholakia
207924d08f test(test_streaming.py): retry if openai is inconsistent with stream options 2024-05-21 18:51:24 -07:00
Krrish Dholakia
f19d7327ca fix(lowest_latency.py): set default none value for time_to_first_token in sync log success event 2024-05-21 18:42:15 -07:00
Ishaan Jaff
67b85d9dda undo ui changes 2024-05-21 18:39:25 -07:00
Krrish Dholakia
20ad1a5189 test(test_router_caching.py): fix caching tests 2024-05-21 18:07:46 -07:00
Ishaan Jaff
7cbdf02a9b ui - set permissions on keys 2024-05-21 18:06:56 -07:00
Krrish Dholakia
bfde1357dd test(test_router_caching.py): fix test 2024-05-21 18:05:05 -07:00
Ishaan Jaff
64130e368c feat - create keys with permissions 2024-05-21 18:04:17 -07:00
Krish Dholakia
aae404fdab
Merge pull request #3771 from BerriAI/litellm_force_anthropic_tool_use
Support anthropic 'tool_choice' param
2024-05-21 18:00:03 -07:00
Krrish Dholakia
af1d209f8f docs(input.md): add anthropic tool choice support to docs 2024-05-21 17:56:21 -07:00
Krrish Dholakia
f3d29a6b4a feat(anthropic.py): support anthropic 'tool_choice' param
Closes https://github.com/BerriAI/litellm/issues/3752
2024-05-21 17:50:44 -07:00