v1.26.6

What's Changed

• fix(llm_guard.py): add streaming hook for moderation calls by @krrishdholakia in #2106

Full Changelog: v1.26.5...v1.26.6

v1.26.5

What's Changed

• fix(sagemaker.py): fix async sagemaker calls by @krrishdholakia in #2103
• fix(presidio_pii_masking.py): enable user to pass their own ad hoc recognizers to presidio by @krrishdholakia in #2100
• (docs) Router - use correct base model for cost by @ishaan-jaff in #2107

Full Changelog: v1.26.4...v1.26.5

v1.26.4

What's Changed

• [FEAT] UI - View all users by @ishaan-jaff in #2101
  

• [FEAT] UI - Create Users on LiteLLM UI by @ishaan-jaff in #2102
  

• [Fix] Unexpected Model Deletion in POST /key/update When Updating Team ID by @ishaan-jaff in #2104

• fix(gemini.py): fix async streaming + add native async completions by @krrishdholakia in #2090

Full Changelog: v1.26.3...v1.26.4

v1.26.3

What's Changed

• feat(llm_guard.py): support llm guard for content moderation by @krrishdholakia in #2087

Full Changelog: v1.26.2...v1.26.3

v1.26.2

What's Changed

• [FEAT] proxy backend - Save User Model Requests in DB by @ishaan-jaff in #2080
• [Feat] UI - allow a user to request access to a model by @ishaan-jaff in #2077
• [FEAT] Admin UI - Approve / Deny user mode requests by @ishaan-jaff in #2089

Full Changelog: v1.26.1...v1.26.2

v1.26.1

What's Changed

• [FEAT] /model/info show models user has access to by @ishaan-jaff in #2066
• [FEAT] UI Set tpm / rpm limits by @ishaan-jaff in #2073

Full Changelog: v1.26.0...v1.26.1

v1.26.0

What's Changed

• feat(llama_guard.py): allow user to define custom unsafe content categories by @krrishdholakia in #2045
• feat(google_text_moderation.py): allow user to use google text moderation for content mod on proxy by @krrishdholakia in #2046
• feat(proxy_server.py): return all teams, user is a member of in /user/info by @krrishdholakia in #2049
• fix(ci): platforms input by @adrien-f in #2060
• fix(ci): set up QEMU and Buildx for multi platform images by @adrien-f in #2065

New Contributors

• @adrien-f made their first contribution in #2060

Full Changelog: v1.25.2...v1.26.0

v1.25.2

Control your companies LLM spend with budgets, call 100+ LLMs

What's Changed

• [FEAT] Track spend per model (for Key, User and Team) by @ishaan-jaff in #2022
• Add safety_settings parameter to gemini generate_content calls by @afbarbaro in #2013
• [FIX] Spend Tracking bug for Keys made on Admin UI (when role = proxy_admin) by @ishaan-jaff in #2047
• feat(presidio_pii_masking.py): allow request level controls for turning on/off pii masking by @krrishdholakia in #2037
• [FEAT] UI- reload key spend info by @ishaan-jaff in https://github.com/BerriAI/lite
  llm/pull/2048
• [FEAT] Budgets per Model (for a key) by @ishaan-jaff in #2029

New Contributors

• @afbarbaro made their first contribution in #2013

Full Changelog: v1.25.1...v1.25.2

v1.25.1

What's Changed

• nit: added urls to pyproject.toml by @ErikBjare in #1536

• ✨ Refresh Helm chart structure by @Lunik in #2018

• [FIX] UI - fix method not allowed error by @ishaan-jaff in #2041

• [FEAT] Set FastAPI root path by @ishaan-jaff in #2039

Important

In a Kubernetes deployment, it's possible to utilize a shared DNS to host multiple applications by modifying the virtual service

👉 Set SERVER_ROOT_PATH in your .env and this will be set as your server root path
Docs here: https://docs.litellm.ai/docs/proxy/deploy#advanced-deployment-settings

Full Changelog: v1.25.0...v1.25.1

v1.25.0

What's Changed

• feat(llama_guard.py): add llama guard support for content moderation + new async_moderation_hook endpoint by @krrishdholakia in #2031
• fix for importllib compatibility issue for python 3.8 by @sorokine in #2017

Full Changelog: v1.24.6...v1.25.0