Please confirm you are human

This browser or connection looks automated. Press and continuously hold the control for 3 seconds to enable Google-hosted web results and, when separately allowed, AI-assisted answers.

A successful check enables 100 search requests. Interactive access does not authorize scraping, systematic collection, or reuse of search output.

Hold with a pointer, or hold Space or Enter.

News

LiteLLM
docs.litellm.ai > docs > providers > valkey_vector_stores

Valkey - Vector Store

1+ day, 6+ hour ago   (874+ words) Search documents you have already indexed in Valkey through LiteLLM's unified vector store API, so that any virtual key can run retrieval against your datastore without ever holding your Valkey credentials. LiteLLM only reads from Valkey; building and loading the…...

LiteLLM
docs.litellm.ai > release_notes > v1.95.0 > v1-95-0

v1.95.0 - Claude Opus 5, MCP Gateway DCR & Rust /v1/messages

2+ week, 2+ day ago   (139+ words) LiteLLM Beyond the new entries, this release raises the advertised context window on azure_ai/claude-opus-4-6, claude-opus-4-7, and claude-opus-4-8 from 200K to 1M, marks the DeepSeek V4 flash and pro families as reasoning models, and records supports_mid_conversation_system across the Claude Fable 5, Sonnet 5, and Opus 4.8 entries on…...

LiteLLM
docs.litellm.ai > release_notes > v1.94.0 > v1-94-0

v1.94.0 - Router Plugins, MCP Client-Held Credentials & Shared DataTable UI

3+ week, 1+ day ago   (196+ words) User budgets now apply on team keys by default. A user's personal max_budget is now enforced whenever they call through a team key, so it stacks with the team and team-member budgets instead of being skipped as before. If you relied…...

LiteLLM
docs.litellm.ai > docs > proxy > guardrails > llm_as_a_judge

LLM-as-a-Judge

3+ week, 4+ day ago   (159+ words) Every judged response costs one extra LLM call to the judge model. Define your guardrails under the guardrails section: Criterion weights must sum to 100. overall_threshold defaults to 80 and on_failure defaults to block. A response that fails the criteria is rejected with HTTP…...

LiteLLM
docs.litellm.ai > release_notes > v1.91.0rc1 > v1-91-0-rc-1

v1.91.0rc1 - MCP OAuth v2, Rust OCR Gateway & Realtime Performance

1+ mon, 2+ week ago   (39+ words) LiteLLM v1.91.0rc1 is the current release candidate for 1.91.0. This release candidate contains changes from existing maintainers only; there are no new contributors in this window....

Google News
docs.litellm.ai > release_notes > v1.86.5 > v1-86-5

v1.86.5 - Claude Fable 5, CrowdStrike AIDR & Batch File Auth

2+ mon, 5+ day ago   (58+ words) LiteLLM v1.86.5 is a patch release on top of v1.86.4. It adds Claude Fable 5, backports the CrowdStrike AIDR identity capture and follow-up fix (an upgrade-monotonicity backport so the 1.86 line keeps AIDR identity attribution), and authorizes batch files using the upload target_model_names....

Google News
docs.litellm.ai > release_notes > v1.87.2 > v1-87-2

v1.87.2 - Claude Fable 5, Batch Auth, CrowdStrike & Mantle SigV4

2+ mon, 5+ day ago   (47+ words) LiteLLM v1.87.2 is a patch release on top of v1.87.1. It adds Claude Fable 5, batch-file authorization, the CrowdStrike AIDR identity pair, and SigV4/IAM auth for the Bedrock Mantle Responses API route....

Google News
docs.litellm.ai > release_notes > v1.86.3 > v1-86-3

v1.86.3 - Gemini 3.5 Flash Day-0 & Pending Line Backports

2+ mon, 5+ day ago   (64+ words) LiteLLM v1.86.3 is a patch release on top of v1.86.2. It closes the gap with the 1.84 and 1.85 lines: day-0 Gemini 3.5 Flash on Vertex AI and Google AI Studio with its paired Vertex tool-call fix, Redis spend-counter seeding, and the observability, budget, and…...

LiteLLM
docs.litellm.ai > docs > auth_overview

Gateway Auth Reference

2+ mon, 1+ week ago   (365+ words) LiteLLM exposes two gateway surfaces that share most authentication and authorization primitives but diverge in a few important places. This page is the side-by-side reference: which header does what, where the two surfaces are symmetric, and where they're not. Each…...

LiteLLM
docs.litellm.ai > docs > search > you_com

You.com Search

2+ mon, 2+ week ago   (104+ words) LiteLLM You.com offers two tiers: If YOUCOM_API_KEY is not set, the adapter automatically uses the keyless endpoint — no signup required to start. With API key (higher limits)​ 2. Start the proxy​ 3. Test the search endpoint​ You.com supports the standard Perplexity…...