httpkobarupayai

QuickBuild

Drop-in OpenAI Chat Completions endpoint: POST {model, messages, max_tokens, temperature, top_p, seed, stream, response_format} and get a chat.completion object (or SSE chunks with stream=true). Models (Cloudflare Workers AI): llama-3.1-8b (default), llama-3.2-3b, llama-3.2-1b, mistral-7b, granite-4.0-micro; any other model name maps to the default (actual model is in the response). Text only, no tool calling; max_tokens capped at 512, input up to 24000 chars. Latest user message checked by Llama Guard 3 (unsafe = 400). Fixed price per call. If the daily capacity is used up you get 503 "capacity". Errors are never charged. GET with ?prompt= also works; with no params it answers a short sample question. Free model list: GET /v1/models.