--- title: "LiteLLM v1.104.6 changes the image-signing key and fixes cost tracking" url: "/release_notes/v1.104.6/v1-104-6" canonical_url: "https://docs.litellm.ai/release_notes/v1.104.6/v1-104-6" type: "release-notes" last_updated: "2026-10-27" summary: "A new cosign key, the eMudhra root CA removed, and fixes for spend on failed and batch requests, token counts and tool calls on gpt-5.4+ models." related: - "/release_notes/v1.104.7/v1-104-7" - "/release_notes/v1.104.5/v1-104-5" --- # LiteLLM v1.104.6 changes the image-signing key and fixes cost tracking > Index of all LiteLLM docs: https://docs.litellm.ai/llms.txt ## Deploy this version **Docker** ```bash docker run \ -e LITELLM_MASTER_KEY=sk- \ -e DATABASE_URL=postgresql://:@:5432/ \ -e STORE_MODEL_IN_DB=True \ -p 4000:4000 \ docker.litellm.ai/berriai/litellm:1.104.6 ``` **Pip** ```bash pip install litellm==1.104.6 ``` This release is published as [`ghcr.io/berriai/litellm:v1.104.6`](https://github.com/BerriAI/litellm/pkgs/container/litellm). See the [GitHub release](https://github.com/BerriAI/litellm/releases/tag/v1.104.6) and the full [releases page](https://github.com/BerriAI/litellm/releases) `v1.104.6` is a patch release on top of [`v1.104.5`](/release_notes/v1.104.5/v1-104-5). It moves image signing to a new cosign key, removes the eMudhra root CA from every certificate bundle, and backports cost tracking, token counting and tool call fixes from `main` onto `stable/1.104.x`. There are no new database migrations. The `v1.104.6` tag points at [`4a8940a`](https://github.com/BerriAI/litellm/commit/4a8940a9fb2696122d683976c389179c92f1a6d2) :::warning The image-signing key has changed Starting with this release, every Docker image and Helm chart is signed with the cosign key from commit [`e82a3c6`](https://github.com/BerriAI/litellm/commit/e82a3c646908ba0a6ee0beddc4310d7a6cac4065). Verifying with the previous key from commit `0112e53` fails. Update anything that pins the old `cosign.pub`, such as `cosign verify` scripts, CI checks and admission policies. ```bash cosign verify \ --key https://raw.githubusercontent.com/BerriAI/litellm/e82a3c646908ba0a6ee0beddc4310d7a6cac4065/cosign.pub \ ghcr.io/berriai/litellm:v1.104.6 ``` ::: ## Key Highlights - **New image-signing key**: Docker images and Helm charts are signed with the cosign key from commit `e82a3c6`, and every certificate bundle drops the eMudhra root CA - **Cost tracking**: failed, timed-out, cancelled and spilled-over requests are priced correctly, and batch cost tracking is fixed - **Tool calls**: gpt-5.4+ models bridged to `/responses` return tool calls on `choices[0]` again ## Image signing and certificates - Every certificate bundle the SDK, proxy and Docker images trust no longer includes the eMudhra root CA - [PR #47670](https://github.com/BerriAI/litellm/pull/47670), [PR #47969](https://github.com/BerriAI/litellm/pull/47969) - Docker images and Helm charts are signed with the cosign key from commit `e82a3c6`. The previous key from commit `0112e53` no longer verifies them - [PR #48103](https://github.com/BerriAI/litellm/pull/48103) ## New Models / Updated Models ### Bug Fixes - **[Azure](../../docs/providers/azure)** - Azure AI Foundry no longer returns 400 when the client sends the `dangerous-tool-use-2026-09-03` beta - [PR #43934](https://github.com/BerriAI/litellm/pull/43934) ## LLM API Endpoints #### Bugs - **Chat completions** - Tool calls from gpt-5.4+ models land on `choices[0]` again instead of `choices[1]`. This started in `v1.97.0` - [PR #46099](https://github.com/BerriAI/litellm/pull/46099) - **Messages API** - `/v1/messages` streams no longer hold `message_start` for minutes during adaptive thinking on Bedrock and Vertex. This started in `v1.100.0` - [PR #43600](https://github.com/BerriAI/litellm/pull/43600) - **Token counting** - DeepSeek and Qwen3.5 token counts on the count endpoints no longer run up to 30% low or twice too high - [PR #47685](https://github.com/BerriAI/litellm/pull/47685) - `/v1/messages/count_tokens` returns HTTP 400 instead of a known-wrong local estimate - [PR #47705](https://github.com/BerriAI/litellm/pull/47705) - **Caching** - Automatic prompt caching no longer adds extra cache marks on `/v1/messages` when `extra_body` overrides a client mark - [PR #43341](https://github.com/BerriAI/litellm/pull/43341) ## Spend Tracking, Budgets and Rate Limiting - The Usage API no longer times out while looking up names for busy hashed keys - [PR #43656](https://github.com/BerriAI/litellm/pull/43656) - Streamed requests through a model alias bill the deployment's price instead of $0. This started in `v1.92.0` - [PR #44341](https://github.com/BerriAI/litellm/pull/44341) - Large non-streaming chat requests to `openai/responses/*` models log and charge spend once instead of twice. This started in `v1.102.0` - [PR #44508](https://github.com/BerriAI/litellm/pull/44508) - Failed, timed-out, cancelled and spilled-over requests are priced correctly instead of wrong or not at all - [PR #47718](https://github.com/BerriAI/litellm/pull/47718) - Batch cost tracking, claude-haiku-5-5 and Bedrock profile pricing, and `safeguards` requests on `/v1/messages` are fixed - [PR #47726](https://github.com/BerriAI/litellm/pull/47726) ## Management Endpoints / UI #### Bugs - Usage reports show the owner of CLI session keys even when spend logs are off or purged - [PR #43642](https://github.com/BerriAI/litellm/pull/43642), [PR #47805](https://github.com/BerriAI/litellm/pull/47805) These fixes and the version bump are backported in [PR #47802](https://github.com/BerriAI/litellm/pull/47802) and [PR #48076](https://github.com/BerriAI/litellm/pull/48076) ### PR roll-up by ownership area - Spend / Budgets / Rate Limits: 5 - LLM API Endpoints: 5 - Other: 3 - UI: 2 - Models & Providers: 1 ## Full Changelog https://github.com/BerriAI/litellm/compare/v1.104.5...v1.104.6 ## Related pages - [v1.104.7](https://docs.litellm.ai/release_notes/v1.104.7/v1-104-7.md) - [v1.104.5](https://docs.litellm.ai/release_notes/v1.104.5/v1-104-5.md)