{"access":{"authentication":{"format":"L402 {macaroon}:{proof_of_payment}","header":"Authorization","protocols":["L402","MPP"],"format_mpp":"Payment <base64url_json>"},"endpoint":"https://sats4ai.com/api/l402/generate-video","method":"POST"},"content_type":"api","description":"Generate cinematic video from a text prompt with MiniMax H3 — top 2 on the Artificial Analysis text-to-video leaderboard — with native audio. Async — returns requestId, poll status at poll_url until COMPLETED.\n\nSend parameters as JSON:\n```\n{\n    \"prompt\": \"A cat walking on the moon\",\n    \"duration\": 5,\n    \"resolution\": \"768p\"\n}\n```\nresolution: \"768p\" or \"2K\" (DEFAULT — omitting resolution and mode yields 2K, as it yielded 1080p before). 2K is upscaled from a 768p render. The retired \"480p\"/\"720p\"/\"1080p\" rungs are still accepted (480p/720p→768p, 1080p→2K), as is legacy \"mode\" standard/pro. Native audio is always on and free — there is no generate_audio flag. duration: 5-15 seconds.\n\nResponse (202): { requestId, jobType, status, poll_url, poll_interval_ms }\nPoll GET poll_url until: { status: \"COMPLETED\", result: { url } }\n\nPricing (per second, by resolution; native audio free): 768p 250 / 2K 400 sats/sec. POST without Authorization for the exact invoice.","name":"Video Generation"}