Free JSON API for LLM data
The same data that powers this site: 164 models across 23 providers, every benchmark score with its source URL, official pricing, and live status from official provider feeds. No API key. CORS enabled. Cached at the edge.
Two lines to your first result
# shell
curl "https://llmmetric.com/v1/models/gpt-6-astra"
# browser / node
const r = await fetch("https://llmmetric.com/v1/compare?a=gpt-6-astra&b=claude-fable-5-1");
const { data } = await r.json();
console.log(data.shared); // [{ benchmark, a: { score, sourceUrl }, b: { ... }, winner }]Endpoints
All GET. OPTIONS preflight supported.| Endpoint | Returns | Query | Cache | Try |
|---|---|---|---|---|
| /v1/models | Every model with specs, pricing and cited benchmark scores. | ?provider=openai&limit=50 | 1h | GET ↗ |
| /v1/models/{id} | One model, plus its provider record. | — | 1h | GET ↗ |
| /v1/benchmarks | Scores grouped by benchmark, sorted by coverage. | ?name=swe | 1h | GET ↗ |
| /v1/pricing | Input, output and cached-input prices per 1M tokens, with source URL. | — | 1h | GET ↗ |
| /v1/status | Current state, 30-day uptime and latency for every probed provider. | — | 60s | GET ↗ |
| /v1/status/{provider} | One provider with daily uptime history and recent incidents. | — | 60s | GET ↗ |
| /v1/compare | Two models side by side on every benchmark they share. | ?a={id}&b={id} | 1h | GET ↗ |
Every response is { data, meta }. meta carries the source, attribution guidance and generatedAt. Lists also include count.
Each benchmark score has sourceUrl and measuredAt; each price has sourceUrl and updatedAt. If we cannot cite it, it is not in the API.
No key is required. Please cite llmmetric.com; upstream records remain subject to their source terms. Cache responses on your side when you can.
Respan's gateway gives you every provider behind one OpenAI-compatible endpoint (https://api.respan.ai/api), with failover, tracing and evals. Free tier, no card.
Methodology and data rules: llmmetric.com/methodology.