What's Changed
- [Feat] Cost Tracking - specify a global vendor discount for costs. by @ishaan-jaff in #15546
- [Feat] UI - Allow setting Provider Discounts on UI by @ishaan-jaff in #15550
- perf(router): use shallow copy instead of deepcopy for model aliases by @AlexsanderHamir in #15576
- perf(router): optimize string concatenation in hash generation by @AlexsanderHamir in #15575
- feat(pricing): Add Gemini 2.5 Flash Image (Nano Banana) in GA by @ndrsfel in #15557
- perf(router): optimize model lookups with O(1) data structures by @AlexsanderHamir in #15578
- Add JP Cross-Region Inference jp.anthropic.claude-haiku-4-5-20251001 by @shinharaguchi in #15598
- fix(cometapi): improve CometAPI provider support (embeddings, image generation, docs) by @TensorNull in #15591
- Fix/entraid app roles jwt claim clean by @TeddyAmkie in #15583
- Add type hint to function_to_dict and fix typo by @aniketmaurya in #15580
- adding support parameters to model_prices json by @FelipeRodriguesGare in #15569
- perf(router): optimize model lookups with O(1) index maps by @AlexsanderHamir in #15574
- Adding new models to the lemonade provider by @eddierichter-amd in #15554
- Prompt caching for anthropic models with openrouter by @konekohana in #15535
- [Fix] CI/CD - Missing env key & Linter type error by @AlexsanderHamir in #15606
- correct claude opus by @Classic298 in #15495
- feature: update pillar security integration to support no persistence mode in litellm proxy by @afogel in #15599
- [Fix] Dall-e-2 for Image Edits API by @ishaan-jaff in #15604
- [OCT] Alexsander PERF improvements by @ishaan-jaff in #15617
- [Feat] UI SSO - Add PKCE for OKTA SSO by @ishaan-jaff in #15608
- [Feat] Allow calling /invoke, /converse routes through AI Gateway + models on config.yaml by @ishaan-jaff in #15618
- [Staging] Oct Staging branch 2 by @ishaan-jaff in #15553
- Litellm staging 10 16 2025 by @krrishdholakia in #15621
- [Fix] Ensure guardrail memory sync after database updates by @orolega in #15633
- fix: Convert object to a correct type by @LucasSugi in #15634
- feat: add guardrail for image generation by @uc4w6c in #15619
- Add support for responses mode in health check by @Sameerlite in #15658
New Contributors
- @shinharaguchi made their first contribution in #15598
- @aniketmaurya made their first contribution in #15580
- @konekohana made their first contribution in #15535
- @afogel made their first contribution in #15599
- @LucasSugi made their first contribution in #15634
Full Changelog: v1.78.2-nightly...v1.78.3-nightly
Docker Run LiteLLM Proxy
docker run \
-e STORE_MODEL_IN_DB=True \
-p 4000:4000 \
ghcr.io/berriai/litellm:main-v1.78.3-nightly
Don't want to maintain your internal proxy? get in touch 🎉
Hosted Proxy Alpha: https://calendly.com/d/4mp-gd3-k5k/litellm-1-1-onboarding-chat