LiteLLM
Named by 18 kept items in this workspace.
12 items name it under frameworks and tools and 6 under classification.technologies.
The items that name it 18
Library with one endpoint to call every model provider. It was compromised again after an API key sat committed on GitHub for five months.
AI gateway library that sits in most agent harness dependency trees (about 92M monthly downloads). Version 1.107.0 was published to PyPI on October 21 at 7:39 AM with a credential-stealing payload and quarantined about 1 hour 43 minutes later. Versions 1.82.7 and 1.82.8 were hit in March by the Team
Supported model endpoint option.
AI gateway (version 1.103.1) compared against Antirion in the benchmark. The page says it fails at rates where Antirion keeps serving.
Model data listing where Fable 5.5 appeared
Example of an OpenAI-compatible provider that can be used as the LLM upstream
Named in Anthropic's report among tools used with proxy farms in distillation campaigns.
Client that can use a room's served API
Open-source (MIT-licensed) AI gateway giving one OpenAI-compatible API to 140+ providers and 1,800+ models, plus agents and MCP servers, with SSO, virtual keys, budgets, rate limits, load balancing, caching, guardrails and logging
An LLM gateway tool that supports automatic routing based on query complexity.
AI gateway (proxy) and Python SDK providing one OpenAI-compatible config routing to 100+ providers, model lists, load-balanced groups, virtual keys/teams/budgets/rate limits, caching, guardrails, observability, and spend tracking
proxy/gateway that speaks the OpenAI-compatible format, one of the providers Graft can be pointed at for its LLM-written summaries
In every workspace you can read 89
Security
AI
All technologies in this workspace.