Skip to main content
LiteLLM is moving to Rust Read the latest updates.

LiteLLM v1.108.12 (LTS) fixes GPT-6.2 caching and Foundry model routers

Released

tip

v1.108.12 is a patch release on the 1.108 LTS line. The lts image tag always points to the latest LTS release, so pull docker.litellm.ai/berriai/litellm:lts to stay on the newest LTS patch. See the LTS pilot.

Deploy this version​

docker run \
-e LITELLM_MASTER_KEY=sk-<paste-a-long-random-key> \
-e DATABASE_URL=postgresql://<user>:<password>@<host>:5432/<dbname> \
-e STORE_MODEL_IN_DB=True \
-p 4000:4000 \
docker.litellm.ai/berriai/litellm:1.108.12

This release is published as ghcr.io/berriai/litellm:v1.108.12. See the GitHub release and the full releases page.

v1.108.12 is a patch release on top of v1.108.11. There are no new database migrations. The v1.108.12 tag points at 5f1b6e4.

What's Changed​

  • OpenAI: Fixed GPT-6.2 prompt caching for system prompts sent to /v1/chat/completions or /v1/messages with store: false - PR #52599 and PR #52694
  • Azure: Fixed azure_ai/model_router on a Foundry /openai/v1 base failing on /v1/responses and /v1/responses/compact - PR #52474
  • Vertex AI and Gemini: Fixed one json_schema row failing every other row in a Vertex batch file - PR #51790
  • Messages API: Fixed 400 Tool reference '<name>' not found on /v1/messages when a guardrail sees a tool name over 64 characters - PR #47434
  • Messages API: Fixed OpenAI models on /v1/messages that keep streaming upstream after the caller closes the stream - PR #52144
  • Guardrails: Fixed tool_policy letting /v1/messages requests through when a blocked tool name is over 64 characters - PR #52717

The changes are backported in PR #52476, PR #52621, PR #52666, PR #52725 and PR #52739.

Full Changelog​

https://github.com/BerriAI/litellm/compare/v1.108.11...v1.108.12

LiteLLM Enterprise
SSO/SAML, audit logs, spend tracking, multi-team management, and guardrails, built for production.
Learn more →