{"openapi":"3.0.3","info":{"title":"TokenMark API","version":"1.0.0","description":"Source-linked local-LLM benchmark configs (tokens per second by model, quant, hardware, backend) and a recommender. Free, no key needed for reads. Same data as https://tokenmark.app/llms.txt."},"servers":[{"url":"https://tokenmark.app"}],"paths":{"/api/recommend":{"get":{"summary":"Recommend models and configs for a hardware target and task list","parameters":[{"name":"hardware","in":"query","schema":{"type":"string"},"description":"e.g. \"Strix Halo\", \"DGX Spark\", \"Mac Ultra\""},{"name":"tasks","in":"query","schema":{"type":"string"},"description":"comma-separated, e.g. coding,reasoning,chat"},{"name":"prefer","in":"query","schema":{"type":"string","enum":["balanced","speed","quality"]}},{"name":"limit","in":"query","schema":{"type":"integer","default":5}}],"responses":{"200":{"description":"recommendations[] with model, vendor, config{quant,backend,variant,hardware,decode_tps,ctx}, reason, source"}}}},"/api/submit":{"post":{"summary":"Queue a GitHub or Hugging Face repo whose README states local-LLM speeds","requestBody":{"required":true,"content":{"application/json":{"schema":{"type":"object","required":["repo"],"properties":{"repo":{"type":"string","example":"owner/name"}}}}}},"responses":{"202":{"description":"queued; poll status_url"}}}},"/api/submit/{id}":{"get":{"summary":"Submission status","parameters":[{"name":"id","in":"path","required":true,"schema":{"type":"string"}}],"responses":{"200":{"description":"status, flags, extracted rows once reviewed"}}}},"/tracker-summary.json":{"get":{"summary":"Full daily snapshot: configs[], platforms, intel, latestReleases","responses":{"200":{"description":"JSON"}}}},"/hardware.json":{"get":{"summary":"Hardware catalogue with official specs per platform","responses":{"200":{"description":"JSON"}}}},"/deployments.json":{"get":{"summary":"Real on-prem local-AI deployments with source articles","responses":{"200":{"description":"JSON"}}}},"/llms-full.txt":{"get":{"summary":"Every single-stream config as one text line per row, grouped by hardware","responses":{"200":{"description":"text/plain"}}}}}}