Open Registry & Telemetry for AI Infrastructure

Discover, compare and monitor LLM models, inference providers, MCP servers and agent skills with source-transparent pricing, context limits, capabilities, access terms and live provider health.

Everything in one place — built on open data, community-maintained.

• Models — canonical registry of LLM models: capabilities, context windows, modalities, licensing, open-weight status, published benchmark results

• Providers — 52 inference providers with API base URLs, auth types, regions, free tiers, trial credits

• Mappings — per-provider pricing, rate limits and model availability; prices are always tied to a provider, never presented as a global model property

• Compare & Calculator — side-by-side model comparison and workload cost estimates based on your own token volumes, cache assumptions and request counts

• Telemetry — provider status and 30-day uptime history from hourly health checks across all 52 providers, plus latency (TTFT p50/p95/p99) monitoring

• MCP Servers — registry of 230 Model Context Protocol servers across 11 categories

• Skills — 260 ready-to-use AI agent prompts and workflows for Claude Code, Cursor, Kiro, Copilot, Windsurf, Gemini CLI, Aider and more

• Insights — data-driven analysis and benchmarks based on live registry and telemetry data

• API, SDK & CLI — public REST API, @openmodels/sdk and a command-line interface for querying the registry

• Multilingual — interface available in English, Chinese, Russian and Spanish

openmodels/     → public YAML registry (models, providers, mappings, schemas)
skills/         → community AI agent skills (github.com/openmodelsrun/skills)
mcp/            → Model Context Protocol server registry
cms/            → Payload CMS for Insights content (cms.openmodels.run)
docs/           → architecture, contributing guides, API reference (docs.openmodels.run)
platform/       → intelligence platform — Next.js + NestJS + Python workers

→ openmodels.run — search, compare, calculator, provider status, skills, MCP servers, insights

The registry is community-maintained. Adding a model, provider, or MCP server is a single YAML file.

→ Contributing Guide