API · Agent files · Data
Developers & AI agents
No key, no auth, CORS wide open. Fetch it from a script, hit it from an agent, or just curl it. Data validated 2026-06-15.
JSON API
Endpoints No auth · CORS open
| GET | /api/index.json | API catalog and discovery links |
| GET | /api/models.json | All 95 models with quant sizes |
| GET | /api/devices.json | All 39 devices with usable memory |
| GET | /api/can-i-run/{model}/{device}.json | Verdict, quant, memory, command for a pair |
Example
$
curl https://localmodel.run/api/can-i-run/llama-3.1-8b/apple-m4-16gb.json Agent-ready endpoints
Discovery files GPTBot · ClaudeBot · PerplexityBot allowed
Every model, device, and compatibility page also serves clean Markdown at its URL plus .md, and exposes these discovery files:
- /llms.txt Concise site map for LLMs
- /llms-full.txt Full dataset as plain text
- /.well-known/agent.json Agent card (capabilities + discovery)
- /sitemap-index.xml Sitemap
Example markdown: /model/llama-3.1-8b.md. AI crawlers (GPTBot, ClaudeBot, PerplexityBot, Google-Extended) are explicitly allowed in robots.txt.
Data is free to use under CC BY 4.0 with attribution to localmodel.run. Figures are estimates; see methodology.