--- title: "v1.90.0 - Six New Providers, OpenTelemetry v2 Parity & Streaming Reliability" url: "/release_notes/v1.90.0/v1-90-0" canonical_url: "https://docs.litellm.ai/release_notes/v1.90.0/v1-90-0" type: "release-notes" last_updated: "2026-10-06" related: - "/release_notes/v1.90.1/v1-90-1" - "/release_notes/v1.89.5/v1-89-5" --- # v1.90.0 - Six New Providers, OpenTelemetry v2 Parity & Streaming Reliability > Index of all LiteLLM docs: https://docs.litellm.ai/llms.txt ## Deploy this version **Docker** ```bash docker run \ -e STORE_MODEL_IN_DB=True \ -p 4000:4000 \ docker.litellm.ai/berriai/litellm:1.90.0 ``` **Pip** ```bash pip install litellm==1.90.0 ``` :::danger[Breaking Changes] **`/key/list` now matches `user_id` and `key_alias` exactly by default.** Earlier releases matched these filters as substrings, so integrations that relied on partial matching will return fewer or different keys after upgrading. The change closes a cross-user key disclosure path. See [PR #30593](https://github.com/BerriAI/litellm/pull/30593). ::: ## Key Highlights - **Six new providers** - ModelScope, LibertAI, Parasail, Pinstripes, TinyFish (search), and FastCRW (search) - plus a new e2b code-execution sandbox primitive. - **91 new models** across Fireworks AI, Scaleway, Tensormesh, LibertAI, Azure AI (including `gpt-5.5` and DeepSeek V4), and Bedrock Mantle. - **OpenTelemetry v2 reaches metrics parity with v1**, emitting the six `gen_ai.client.*` metrics, stamping input/output message content, and scoping OTLP credentials per tenant. - **A broad streaming-reliability sweep**: upstream connections are now released when the client disconnects mid-stream (Gemini, aiohttp), requests are cancelled cleanly, and partial spend is recorded on interrupted streams. - **Two new guardrails** (Cisco AI Defense, Repello Argus) and a large Next.js App Router UI migration covering the models, teams, users, organizations, api-keys, and usage pages. ## App Router Routing We're moving the Admin UI from query param based routing to the Nextjs App Router. The motivation is that routing now lives in the URL, so any view (a specific team, a filtered usage report, a single key) becomes a shareable link you can send to a teammate or bookmark instead of a piece of in-memory client state. The motivation for this is twofold: it lays the groundwork for a lot of highly requested feaetures/improvements to the UI, and it lays the groundwork for contributing to LiteLLM easier and code more human reviewable for maintainers. The biggest one here will be sharing links for different pages like specific logs pages, teams pages, and more. ## New Providers and Endpoints ### New Providers (6 new providers) | Provider | Supported LiteLLM Endpoints | Description | | --- | --- | --- | | ModelScope (`modelscope`) | Chat Completions | OpenAI-compatible provider for ModelScope-hosted models - [PR #28460](https://github.com/BerriAI/litellm/pull/28460) | | LibertAI (`libertai`) | Chat Completions, Embeddings | JSON-configured OpenAI-compatible provider; ships 12 catalog models including `bge-m3` embeddings - [PR #30203](https://github.com/BerriAI/litellm/pull/30203) | | TinyFish (`tinyfish`) | Search | Web search provider - [PR #30634](https://github.com/BerriAI/litellm/pull/30634) | | FastCRW (`fastcrw`) | Search | Web search provider - [PR #30434](https://github.com/BerriAI/litellm/pull/30434) | | Parasail (`parasail`) | Chat Completions | OpenAI-compatible provider | | Pinstripes (`pinstripes`) | Chat Completions | New chat provider; ships 6 catalog models | ### New LLM API Endpoints | Capability | Description | Documentation | | --- | --- | --- | | Code execution (e2b) | New sandbox / code-interpreter primitive for running model-generated code - [PR #30898](https://github.com/BerriAI/litellm/pull/30898) | [Sandbox](../../docs/sandbox) | ## New Models / Updated Models #### New Model Support (91 new models) | Provider | Model | Context | Input ($/1M) | Output ($/1M) | Features | | --- | --- | --- | --- | --- | --- | | Azure AI | `azure_ai/gpt-5.5` | 1,050,000 | $5 | $30 | reasoning, function calling, prompt caching, pdf, vision | | Azure AI | `azure_ai/gpt-5.5-2026-04-23` | 1,050,000 | $5 | $30 | reasoning, function calling, prompt caching, pdf, vision | | Azure AI | `azure_ai/deepseek-v4-flash` | 1,000,000 | $0.19 | $0.51 | reasoning, function calling | | Azure AI | `azure_ai/deepseek-v4-pro` | 1,000,000 | $1.74 | $3.48 | reasoning, function calling | | Azure AI | `azure_ai/deepseek-v3.1` | 131,072 | $1.23 | $4.94 | reasoning, function calling | | Azure AI | `azure_ai/MAI-Image-2.5` | - | $5 | - | image generation | | Azure AI | `azure_ai/MAI-Image-2.5-Flash` | - | $1.75 | - | image generation | | Azure AI | `azure_ai/MAI-Image-2e` | - | $5 | - | image generation | | Azure | `azure/gpt-realtime-whisper` | - | - | - | audio transcription | | OpenAI | `gpt-realtime-whisper` | - | - | - | audio transcription | | DeepSeek | `deepseek-v4-flash` / `deepseek/deepseek-v4-flash` | 1,000,000 | $0.14 | $0.28 | function calling, prompt caching | | DeepSeek | `deepseek-v4-pro` / `deepseek/deepseek-v4-pro` | 1,000,000 | $0.43 | $0.87 | function calling, prompt caching | | Mistral | `mistral/mistral-medium-3-5` | 262,144 | $1.50 | $7.50 | function calling, vision | | GitHub Copilot | `github_copilot/mai-code-1-flash` | 128,000 | $0.75 | $4.50 | function calling | | Fireworks AI | 24 models incl. `deepseek-v4-pro`, `glm-5p2`, `kimi-k2p6`/`kimi-k2p7-code`, `minimax-m3`, `qwen3p7-plus`, `gpt-oss-120b`/`gpt-oss-20b` | up to 1,048,576 | $0.07-$2.80 | $0.28-$8.80 | function calling, reasoning, vision | | Bedrock Mantle | `bedrock_mantle/google.gemma-4-26b-a4b` / `gemma-4-31b` / `gemma-4-e2b` | 128k-256k | $0.04-$0.14 | $0.08-$0.40 | function calling, reasoning, vision | | LibertAI | 12 models incl. `qwen3.6-35b-a3b(-thinking)`, `gemma-4-31b-it(-thinking)`, `deepseek-v4-flash`, `bge-m3` | up to 262,144 | $0.01-$0.25 | free-$1.75 | function calling, reasoning, vision, embedding | | Pinstripes | 6 models incl. `ps/minimax-m2.7`, `ps/qwen3.6-35b-a3b`, `ps/glm-4.5-air`, `ps/deepseek-v4-flash` | up to 1,000,192 | $0.09-$0.30 | $0.20-$0.60 | function calling, reasoning | | Scaleway | 17 models incl. `qwen3.5-397b-a17b`, `mistral-medium-3.5-128b`, `gemma-4-26b-a4b-it`, `gpt-oss-120b`, `whisper-large-v3` | up to 256,000 | free-$1.50 | free-$7.50 | function calling, reasoning, vision, audio, embedding | | Tensormesh | 10 models incl. `Qwen3-Coder-480B-A35B-FP8`, `Qwen3.5-397B-A17B-FP8`, `Kimi-K2.6`, `DeepSeek-V4-Flash`, `gpt-oss-120b`/`gpt-oss-20b` | up to 262,144 | $0.07-$1.40 | $0.28-$4.40 | function calling, reasoning, prompt caching | | Soniox | `soniox/stt-async-v5` | 8,000 | - | - | audio transcription | | TinyFish | `tinyfish/search` | - | - | - | search | The 91 new entries also include the full `fireworks_ai/accounts/...` model and router paths. Claude Fable 5 already shipped in v1.89.0, so it is not counted here. Full diff: `model_prices_and_context_window.json`. #### Features - **[Anthropic](../../docs/providers/anthropic)** - Surface compaction usage iterations data - [PR #27065](https://github.com/BerriAI/litellm/pull/27065) - Serve Anthropic-native `/v1/models` for Claude Code gateway discovery - [PR #30273](https://github.com/BerriAI/litellm/pull/30273) - **[OpenRouter](../../docs/providers/openrouter)** - Map reasoning `max` level to `xhigh` - [PR #28881](https://github.com/BerriAI/litellm/pull/28881) - **[Bedrock](../../docs/providers/bedrock)** - Optionally forward multimodal content blocks in AgentCore `InvokeAgentRuntime` - [PR #28885](https://github.com/BerriAI/litellm/pull/28885) - Support file content retrieval for batch output files - [PR #30595](https://github.com/BerriAI/litellm/pull/30595) - Make Bedrock Mantle Responses routing data-driven per model - [PR #30700](https://github.com/BerriAI/litellm/pull/30700) - **[DashScope](../../docs/providers/dashscope)** - Add Responses API support - [PR #30286](https://github.com/BerriAI/litellm/pull/30286) - **[OCI](../../docs/providers/oci)** - Make Cohere `{{trace}}` judges work (tool param types + agentic tool-calling continuation) - [PR #30646](https://github.com/BerriAI/litellm/pull/30646) #### Bug Fixes - **[Anthropic](../../docs/providers/anthropic)** - Apply `cache_control_injection_points` on the `/v1/messages` path - [PR #30341](https://github.com/BerriAI/litellm/pull/30341) - Strip LiteLLM-injected `total_tokens` from `/v1/messages` responses - [PR #30382](https://github.com/BerriAI/litellm/pull/30382) - Cap cache_control injection at 4 blocks - [PR #30480](https://github.com/BerriAI/litellm/pull/30480) - Drop orphaned `server_tool_use` on multi-turn replay from generic OpenAI clients - [PR #30486](https://github.com/BerriAI/litellm/pull/30486) - Don't leak tool `type` into OpenAI function parameters schema - [PR #30618](https://github.com/BerriAI/litellm/pull/30618) - **[Bedrock](../../docs/providers/bedrock)** - Preserve `cache_control` for ARN models in the `/v1/messages` adapter - [PR #29823](https://github.com/BerriAI/litellm/pull/29823) - Handle `role: "system"` inside the messages array on `/v1/messages` - [PR #30443](https://github.com/BerriAI/litellm/pull/30443) - Use a unique function-call id for Bedrock Mantle responses->chat tool calls - [PR #30426](https://github.com/BerriAI/litellm/pull/30426) - Add SigV4 fallback to Bedrock Mantle chat completions auth - [PR #30714](https://github.com/BerriAI/litellm/pull/30714) - **[Gemini / Vertex AI](../../docs/providers/gemini)** - Use `get_vertex_base_url` for `cachedContents` host - [PR #29707](https://github.com/BerriAI/litellm/pull/29707) - Buffer native Gemini SSE frames - [PR #30225](https://github.com/BerriAI/litellm/pull/30225) - Map Gemini upstream-error body code 429 to `RateLimitError` - [PR #30417](https://github.com/BerriAI/litellm/pull/30417) - Ensure checks show `gemini-3-flash-preview` supports `responseJsonSchema` - [PR #30696](https://github.com/BerriAI/litellm/pull/30696) - **[OpenAI-compatible](../../docs/providers/openai_compatible)** - Preserve `cache_control` for OpenAI-compatible custom endpoints - [PR #30387](https://github.com/BerriAI/litellm/pull/30387) - hosted_vllm: remove `thinking_blocks` and convert list content to strings - [PR #30475](https://github.com/BerriAI/litellm/pull/30475) - Don't stack provider prefix on wildcard models with a custom prefix - [PR #30360](https://github.com/BerriAI/litellm/pull/30360) - **[WatsonX](../../docs/providers/watsonx)** - Wrap string embedding input in an array for the WatsonX API - [PR #30897](https://github.com/BerriAI/litellm/pull/30897) - **Pricing / Cost map** - Add cost mapping for `deepseek-v4-flash`/`deepseek-v4-pro` - [PR #27056](https://github.com/BerriAI/litellm/pull/27056) - Add `mistral-medium-3-5` to the cost map - [PR #29303](https://github.com/BerriAI/litellm/pull/29303) - Add `azure_ai/gpt-5.5` to the model cost map - [PR #30428](https://github.com/BerriAI/litellm/pull/30428) - Add GitHub Copilot MAI Code Flash pricing - [PR #30415](https://github.com/BerriAI/litellm/pull/30415) - Sync the Fireworks AI model registry with the current platform catalog - [PR #30616](https://github.com/BerriAI/litellm/pull/30616) - Add `soniox/stt-async-v5` - [PR #30672](https://github.com/BerriAI/litellm/pull/30672) - Correct swapped input/output token costs for `command-r7b-12-2024` - [PR #30413](https://github.com/BerriAI/litellm/pull/30413) - Add 1h cache-write cost for Anthropic Sonnet 4.5/4.6 - [PR #30474](https://github.com/BerriAI/litellm/pull/30474) - Route Volcengine (Doubao) tiered-pricing models to the tiered cost handler - [PR #30357](https://github.com/BerriAI/litellm/pull/30357); sort tiered thresholds numerically - [PR #30375](https://github.com/BerriAI/litellm/pull/30375); treat a DashScope explicit `0.0` tier cost as a real price - [PR #30653](https://github.com/BerriAI/litellm/pull/30653) - Drop synthesized zero costs in `register_model` to preserve sparse entries - [PR #30201](https://github.com/BerriAI/litellm/pull/30201) ## LLM API Endpoints #### Features - **[Responses API](../../docs/response_api)** - Propagate `completed_response` through `FallbackResponsesStreamWrapper` for streaming `/v1/responses` container ownership - [PR #30213](https://github.com/BerriAI/litellm/pull/30213) - **[/v1/models](../../docs/proxy/model_management)** - Surface `max_input_tokens`/`max_output_tokens` on `/v1/models` - [PR #30272](https://github.com/BerriAI/litellm/pull/30272) - Include model group aliases in v1 model info - [PR #30626](https://github.com/BerriAI/litellm/pull/30626) - **[Realtime](../../docs/realtime)** - Allow non-admin virtual keys to call GA Realtime WebRTC HTTP routes - [PR #30089](https://github.com/BerriAI/litellm/pull/30089) - **[Files](../../docs/files_endpoints)** - Attach existing OpenAI file ids - [PR #30628](https://github.com/BerriAI/litellm/pull/30628) #### Bugs - **General** - Token counter: handle Anthropic `tool_reference` blocks to stop dropped spend logs - [PR #30302](https://github.com/BerriAI/litellm/pull/30302) - Streaming: guard `raise_on_model_repetition` against empty choices - [PR #30485](https://github.com/BerriAI/litellm/pull/30485) - Audio: don't override an explicit `response_format` with `verbose_json` - [PR #30599](https://github.com/BerriAI/litellm/pull/30599) - Validate the resolved model in `/realtime/client_secrets` for non-transcription sessions - [PR #30710](https://github.com/BerriAI/litellm/pull/30710) ## Management Endpoints / UI #### Features - **App Router migration** - models - [PR #30677](https://github.com/BerriAI/litellm/pull/30677), teams - [PR #30343](https://github.com/BerriAI/litellm/pull/30343), users - [PR #30334](https://github.com/BerriAI/litellm/pull/30334), organizations - [PR #30336](https://github.com/BerriAI/litellm/pull/30336), api-keys - [PR #30699](https://github.com/BerriAI/litellm/pull/30699), usage report - [PR #30694](https://github.com/BerriAI/litellm/pull/30694), agents + router-settings - [PR #30323](https://github.com/BerriAI/litellm/pull/30323) - **UI cleanup** - remove the unreachable `/chat` page - [PR #30178](https://github.com/BerriAI/litellm/pull/30178), dead UI components - [PR #30340](https://github.com/BerriAI/litellm/pull/30340), orphaned pass-through-settings route - [PR #30692](https://github.com/BerriAI/litellm/pull/30692); remove in-product survey and feedback nudges - [PR #30773](https://github.com/BerriAI/litellm/pull/30773) - **Virtual Keys** - expose per-model budget usage in `/key/info` - [PR #30394](https://github.com/BerriAI/litellm/pull/30394); grace-period key rotation returns the deprecated-key lookup result on 401 - [PR #30327](https://github.com/BerriAI/litellm/pull/30327) - **Teams / Orgs** - add `key_limit` query param to `/team/info` - [PR #30006](https://github.com/BerriAI/litellm/pull/30006); list public team model names in `/v1/models` - [PR #30588](https://github.com/BerriAI/litellm/pull/30588) - **Proxy CLI Auth** - add `verification_uri_complete` to the CLI SSO device flow - [PR #30571](https://github.com/BerriAI/litellm/pull/30571) - **Proxy** - configurable response headers and login-page hint - [PR #30792](https://github.com/BerriAI/litellm/pull/30792); gate the "Default Credentials" hint on `/ui/login` behind an env flag - [PR #30234](https://github.com/BerriAI/litellm/pull/30234) #### Bugs - **Access control / keys** - `/key/list` now does exact `user_id`/`key_alias` matching by default, preventing cross-user key disclosure - [PR #30593](https://github.com/BerriAI/litellm/pull/30593) - Restrict `/customer/daily/activity` to admin-only - [PR #28849](https://github.com/BerriAI/litellm/pull/28849) - `org_admin` sees all org teams when the UI sends its own `user_id` - [PR #30247](https://github.com/BerriAI/litellm/pull/30247) - Allow internal roles to access vector store CRUD routes - [PR #30503](https://github.com/BerriAI/litellm/pull/30503) - Require premium only when enabling premium metadata fields - [PR #30506](https://github.com/BerriAI/litellm/pull/30506) - Guard `check_and_fix_namespace` against a `None` key - [PR #30435](https://github.com/BerriAI/litellm/pull/30435) - Warn at startup when `custom_auth` skips `common_checks` enforcement - [PR #30665](https://github.com/BerriAI/litellm/pull/30665) - Resolve list-files credentials from team BYOK deployments - [PR #30495](https://github.com/BerriAI/litellm/pull/30495); preserve `azure_ad_token` through `CredentialLiteLLMParams` for `/v1/files` + batches - [PR #30241](https://github.com/BerriAI/litellm/pull/30241) - Enforce budget for models not in the cost map - [PR #24949](https://github.com/BerriAI/litellm/pull/24949) - **UI** - Stop the Virtual Keys page from an infinite render loop - [PR #30397](https://github.com/BerriAI/litellm/pull/30397) - Source api-keys identity from `useAuthorized` to stop "User ID is not set" - [PR #30903](https://github.com/BerriAI/litellm/pull/30903) - Render logos correctly under a custom `server_root_path` - [PR #31156](https://github.com/BerriAI/litellm/pull/31156) - Warn that team models are deleted in the delete-team modal - [PR #29990](https://github.com/BerriAI/litellm/pull/29990) - Three small fixes - Gemini `api_base`, credential form reset, Mode badge - [PR #30419](https://github.com/BerriAI/litellm/pull/30419) - Repoint the dead usage-guide link to cost-tracking docs - [PR #30859](https://github.com/BerriAI/litellm/pull/30859) - **Proxy** - Support SMTP implicit SSL (port 465) - [PR #30395](https://github.com/BerriAI/litellm/pull/30395) ## AI Integrations ### Logging - **[OpenTelemetry](../../docs/proxy/logging)** - Emit the six `gen_ai.client.*` metrics at v1 parity in v2 - [PR #30326](https://github.com/BerriAI/litellm/pull/30326) - One v2 logger owns the global provider; scope tenant OTLP creds per exporter - [PR #30590](https://github.com/BerriAI/litellm/pull/30590) - Export v2 gen_ai client metrics to the configured meter provider - [PR #30549](https://github.com/BerriAI/litellm/pull/30549) - Stamp `gen_ai.input/output.messages` on v2 spans - [PR #30548](https://github.com/BerriAI/litellm/pull/30548) - Cap metric attribute cardinality with include/exclude lists - [PR #30257](https://github.com/BerriAI/litellm/pull/30257) - Record the full error message on the standard exception event in v2 - [PR #30380](https://github.com/BerriAI/litellm/pull/30380) - Accept `UPPER_SNAKE_CASE` `OTEL_INSTRUMENTATION_GENAI_CAPTURE_MESSAGE_CONTENT` in v2 - [PR #30562](https://github.com/BerriAI/litellm/pull/30562) - **General** - Preserve `error_message` on `ProxyException` failures in spend logs - [PR #30381](https://github.com/BerriAI/litellm/pull/30381) ### Guardrails - **Cisco AI Defense** - new integration - [PR #28249](https://github.com/BerriAI/litellm/pull/28249) - **Repello Argus** - new integration - [PR #30465](https://github.com/BerriAI/litellm/pull/30465) - **[Presidio](../../docs/proxy/guardrails/pii_masking_v2)** - add missing UK PII entity types - [PR #30537](https://github.com/BerriAI/litellm/pull/30537); don't mask the live request when the guardrail is `logging_only` - [PR #30461](https://github.com/BerriAI/litellm/pull/30461) - **AIM** - return 400 not 500 when AIM blocks a request - [PR #30573](https://github.com/BerriAI/litellm/pull/30573) - **General** - Stop re-initializing DB guardrails on every poll - [PR #30542](https://github.com/BerriAI/litellm/pull/30542) - Run the `pre_call` hook once for model-level guardrails - [PR #30543](https://github.com/BerriAI/litellm/pull/30543) - `disable_global_guardrails` overrides the team list - [PR #28563](https://github.com/BerriAI/litellm/pull/28563) - Surface OpenAI moderation `violation_categories` on guardrail traces - [PR #30659](https://github.com/BerriAI/litellm/pull/30659) ### Secret Managers - **[AWS Secrets Manager](../../docs/secret)** - cross-region replication - [PR #30368](https://github.com/BerriAI/litellm/pull/30368) ## Spend Tracking, Budgets and Rate Limiting - **Service-tier pricing** - apply the `service_tier` suffix to above-threshold cache rates and expose priority+threshold keys in `ModelInfo` - [PR #30450](https://github.com/BerriAI/litellm/pull/30450); price and surface the Anthropic response `service_tier` in cost tracking - [PR #30558](https://github.com/BerriAI/litellm/pull/30558); stop non-string `service_tier` from silently dropping cost tracking - [PR #30690](https://github.com/BerriAI/litellm/pull/30690), [PR #30706](https://github.com/BerriAI/litellm/pull/30706) - **Budgets** - enforce budgets against authoritative DB spend when the cross-pod counter is stale - [PR #30684](https://github.com/BerriAI/litellm/pull/30684); release a budget reservation when a request is cancelled mid-flight - [PR #30522](https://github.com/BerriAI/litellm/pull/30522); recompute `budget_reset_at` when `budget_duration` changes - [PR #30555](https://github.com/BerriAI/litellm/pull/30555) - **Rate limiting** - prevent internal `parallel_request_limiter` fields from leaking to upstream providers - [PR #30545](https://github.com/BerriAI/litellm/pull/30545) - **Spend accuracy** - record partial spend on the failure row for interrupted streams - [PR #30788](https://github.com/BerriAI/litellm/pull/30788); recover output tokens for interrupted Anthropic streams - [PR #30787](https://github.com/BerriAI/litellm/pull/30787); stop Perplexity double-billing reasoning tokens in the manual cost fallback - [PR #30488](https://github.com/BerriAI/litellm/pull/30488); correct cached-token usage with `ChatCompletionUsageBlock` - [PR #30422](https://github.com/BerriAI/litellm/pull/30422) - **Usage aggregation** - drain all daily-spend batches per flush cycle - [PR #30505](https://github.com/BerriAI/litellm/pull/30505); show session-aggregate cost and duration in request logs - [PR #30507](https://github.com/BerriAI/litellm/pull/30507); coalesce null aggregates for no-spend keys - [PR #29945](https://github.com/BerriAI/litellm/pull/29945); remove timezone date expansion in daily-activity aggregation - [PR #29569](https://github.com/BerriAI/litellm/pull/29569) ## MCP Gateway - Make the MCP gateway name and description configurable via env vars - [PR #30473](https://github.com/BerriAI/litellm/pull/30473) - Fail closed when the scope filter resolves to no servers - [PR #30353](https://github.com/BerriAI/litellm/pull/30353) - Re-raise instead of silently dropping MCP team permissions - [PR #30477](https://github.com/BerriAI/litellm/pull/30477) - Drop the phantom 401 span on delegated OAuth2 tool calls - [PR #30494](https://github.com/BerriAI/litellm/pull/30494) - Challenge delegate-auth OAuth servers with the upstream `resource_metadata` - [PR #31255](https://github.com/BerriAI/litellm/pull/31255) - Default the Linear MCP registry entry to streamable HTTP - [PR #30396](https://github.com/BerriAI/litellm/pull/30396) - Preserve native tools in the semantic filter hook - [PR #26650](https://github.com/BerriAI/litellm/pull/26650) ## Performance / Loadbalancing / Reliability improvements - **Streaming connection hygiene** - cancel the upstream Gemini request and release the httpx connection on client disconnect - [PR #30075](https://github.com/BerriAI/litellm/pull/30075); close the upstream LLM stream when the client disconnects mid-stream - [PR #30245](https://github.com/BerriAI/litellm/pull/30245); release the aiohttp connection when stream iteration ends abnormally - [PR #30271](https://github.com/BerriAI/litellm/pull/30271); use `e.request_data` for `logging_obj` in `ModifyResponseException` streaming passthrough - [PR #30800](https://github.com/BerriAI/litellm/pull/30800) - **Caching** - add a valkey-semantic cache backend and fix semantic-cache scope keys - [PR #30675](https://github.com/BerriAI/litellm/pull/30675); url-encode the object name in the GCS cache GET path - [PR #30378](https://github.com/BerriAI/litellm/pull/30378); allow `use_redis_transaction_buffer` without a Redis cache - [PR #28764](https://github.com/BerriAI/litellm/pull/28764) - **Router / fallbacks** - resolve a list-unhashable crash on model alias - [PR #30464](https://github.com/BerriAI/litellm/pull/30464); clean pattern_router state on upsert/delete - [PR #29601](https://github.com/BerriAI/litellm/pull/29601); preserve the fallback model in SDK fallback responses - [PR #28260](https://github.com/BerriAI/litellm/pull/28260); add `expose_router_debug_in_errors` (default True) to redact internal model_group/fallback names - [PR #30418](https://github.com/BerriAI/litellm/pull/30418) - **Startup / workers** - fail fast on a non-PostgreSQL `DATABASE_URL` instead of hanging - [PR #30366](https://github.com/BerriAI/litellm/pull/30366); add `--max_requests_before_restart_jitter` to stagger worker restarts - [PR #30601](https://github.com/BerriAI/litellm/pull/30601); fix the IAM refresh-engine watcher race - [PR #30183](https://github.com/BerriAI/litellm/pull/30183); release the cron pod-lock by matching `async_set_cache` JSON encoding - [PR #30600](https://github.com/BerriAI/litellm/pull/30600) - **Health checks** - correct Bedrock embedding health checks - [PR #30583](https://github.com/BerriAI/litellm/pull/30583); bump the health-check `max_tokens` default to 16 for GPT-5 compatibility - [PR #30708](https://github.com/BerriAI/litellm/pull/30708), [PR #26610](https://github.com/BerriAI/litellm/pull/26610) - **Developer experience / CI** - around 30 PRs hardening the lint and type-check gates (standardizing on basedpyright, dropping mypy, ratcheting any-discipline budgets), an osv-scanner lockfile workflow, zizmor PR gating, a local fake-OpenAI test endpoint replacing the shared mock, dependency bumps, and a pinned build toolchain. ## Documentation Updates - Add 1-click AWS/GCP Terraform deploy buttons and fix README deploy-button rendering - [PR #29879](https://github.com/BerriAI/litellm/pull/29879) - Strengthen the coding conventions in `CLAUDE.md` - [PR #30333](https://github.com/BerriAI/litellm/pull/30333) - Clarify the Linear portion of the PR template - [PR #30766](https://github.com/BerriAI/litellm/pull/30766) ## New Contributors @hannahmadison, @ayushh0110, @Dotify71, @munnr, @V-3604, @yrk111222, @Silvenga, @djmaze, @apshada, @HumphreySun98, @Harshxth, @tomoyat1, @S0ngRu1, @habonlaci, @moshemalawach, @nahrinoda, @Vedant-Agarwal, @lollinng, @anneheartrecord, @hdt12a1, @vineethsaivs, @krishvsoni, @rvishwas26, @santino18727-debug, @darktheorys, @songkuan-zheng, @Thijmen, @Kropiunig, @jay-tau, @KnyazSh, @koztkozt, @us, @Anuj7411, @zkryakgul, @lavish619, @EugeneLugovtsov, @Bochenski, @menardorama, @factnn, @semmons99, @nitishagar, @FadelT, @jho1-godaddy, @yucheng-berri, @ad1269, @shzdehmd, @vanika02, @Nithish-Yenaganti, @simantak-dabhade, @devYRPauli, @clpatterson, @tcconnally ## Full Changelog [`v1.89.0...v1.90.0`](https://github.com/BerriAI/litellm/compare/v1.89.0...v1.90.0) ## Related pages - [v1.90.1](https://docs.litellm.ai/release_notes/v1.90.1/v1-90-1.md) - [v1.89.5](https://docs.litellm.ai/release_notes/v1.89.5/v1-89-5.md)