{"help":["fitllm agent API — will this LLM fit this hardware? (no auth, no key)","","GET /api/check?model=<name>&gpu=<name>        GPU fit (Q4_K_M, 8K ctx defaults)","GET /api/check?model=<name>&gpu=5090+3090     multi-GPU rig (VRAM pools, mixed cards OK)","GET /api/check?model=<name>&gpu=3090&count=2  N identical cards","GET /api/check?model=<name>&ram=64            Apple Silicon unified memory (GB)","","optional: &quant=Q4_K_M|Q5_K_M|Q6_K|Q8_0|FP16 (GPU) | 4|8|16 (Mac bits)","          &ctx=<tokens>  &kv=16|8|4  &format=json|text (default: json; curl gets text)","","examples:","  curl 'https://fitllm.run/api/check?model=gemma%204%2031b&gpu=4090'","  curl 'https://fitllm.run/api/check?model=qwen%203.6%2035b&gpu=5090%2B3090&ctx=32768'","","exit-code guard for scripts: npx fitllm \"<model>\" --gpu \"<gpu>\"  (exit 0/1)","data: https://fitllm.run/data (census CSV/JSON, CC0) · engine: github.com/click6067-ship-it/fitllm-engine (MIT)","docs: this page · MCP server: https://fitllm.run/api/mcp"]}