Data productv1 public, read-only

LLMReference API

v1 is a read-only public JSON API and MCP server for the same model, provider, pricing, and benchmark catalog that powers this site. There is no authentication and there are no accounts.

Catalog counts

2,320

Public models

177

Providers

3,103

Routes

203

Benchmarks

Endpoints

MethodPathReturns
GET/api/v1API index
GET/api/v1/modelsList models (filters, sort, pagination)
GET/api/v1/models/{slug}One model: specs, capabilities, context, freshness, lifecycle
GET/api/v1/models/{slug}/routesProvider routes and pricing for one model
GET/api/v1/models/{slug}/scoresBenchmark scores for one model
GET/api/v1/providersList providers
GET/api/v1/providers/{slug}One provider
GET/api/v1/providers/{slug}/routesRoutes and pricing for one provider
GET/api/v1/benchmarksList benchmarks
GET/api/v1/benchmarks/{slug}One benchmark
GET/api/v1/benchmarks/{slug}/scoresScores for one benchmark
POST/mcpStreamable HTTP MCP (JSON-RPC)

GET /api/v1/models parameters

ParamNotes
providerProvider slug (exact). Restricts models that have a route on that provider.
modalitytext, image, video, audio, or multimodal.
open-weightstrue or false. Alias: openWeights. True includes open-source and open-weights models.
opennessopen-source, open-weights, proprietary, or unknown.
familyFamily slug (exact).
categoryMatches specialization or a use-case tag.
specializationExact specialization string.
deprecatedtrue or false.
sortslug (default), name, release, or lastResearched.
orderasc (default) or desc.
limitPage size. Default 50, max 200.
offsetNumber of matches to skip. Default 0.
  • Success bodies are { "data": ..., "meta": { "version": "v1", "generatedAt", "dataSha", "count", "total?", "limit?", "offset?" } }.
  • Unknown slugs return 404 JSON. Invalid query parameters return 400 JSON.
  • Score rows include selfReported and evidence (self-reported or confirmed).
  • Route objects include pricingAvailability (for example listed or unpublished).
  • CORS is *, including OPTIONS preflight. Cache-Control is public, max-age=0, must-revalidate (catalog changes on deploy). x-llmref-data-sha is also set.
  • Rate limit is a best-effort per-IP cap of 60 requests / 60 seconds on /api/v1 and /mcp.

curl examples

List models

curl -sS 'https://www.llmreference.com/api/v1/models?provider=openai-api&modality=text&limit=5'

One model

curl -sS 'https://www.llmreference.com/api/v1/models/gpt-5'

Model routes

curl -sS 'https://www.llmreference.com/api/v1/models/gpt-5/routes'

Model scores

curl -sS 'https://www.llmreference.com/api/v1/models/gpt-5/scores'

Providers

curl -sS 'https://www.llmreference.com/api/v1/providers?limit=10'

Provider routes

curl -sS 'https://www.llmreference.com/api/v1/providers/openai-api/routes'

Benchmarks

curl -sS 'https://www.llmreference.com/api/v1/benchmarks?limit=10'

MCP initialize

curl -sS -X POST 'https://www.llmreference.com/mcp' -H 'Content-Type: application/json' -H 'Accept: application/json, text/event-stream' -d '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2025-06-18","capabilities":{},"clientInfo":{"name":"curl","version":"0.0.1"}}}'

MCP client config

POST JSON-RPC to /mcp (Streamable HTTP). Tools: search_models, get_model, list_provider_routes, get_scores.

{
  "mcpServers": {
    "llmreference": {
      "url": "https://www.llmreference.com/mcp"
    }
  }
}

Last refreshed 2026-11-14. Next refresh: weekly.

Freshness on this page is the latest last-researched date in the seed catalog that backs GET /api/v1.