Skip to main content
2026/8/23 22:40 (UTC+8) · Docs Update 🚀 13 image and video model pages now carry an AI Agent integration prompt — copy one block and let your coding agent do the integration or the triage Each model’s overview page now has a one-click copyable prompt just below the intro. Paste it into Codex, Claude Code, Cursor or any coding agent, and it first fetches the plain-text version of that page (append .md to any docs URL), then writes code in your project’s own stack rather than copying the doc samples verbatim. The prompt hard-codes the integration red lines people trip over most:
  • Timeout values: image and video latencies differ widely, and a default 30-60 second client timeout cuts off perfectly healthy requests — which are still billed
  • How to read the response: image models mostly return base64, while FLUX returns a URL that dies in about 10 minutes; video is async polling plus a 24-hour signed link, and both need re-hosting immediately
  • Compress before upload: one standard of 2048px long edge at quality 0.9, so raw phone photos never go straight up
  • Per-model red lines: never send auto to gpt-image-2, never send generateAudio to VEO, never route Wan through /v1/videos
Coverage spans the three gpt-image-2 variants, Nano Banana Pro / 2 / Lite, FLUX, Seedream and Grok Imagine on the image side, plus Seedance 2.0, Wan2.7, HappyHorse and VEO 3.1 Official on the video side. Use it to get a new integration running in one pass, or as a checkup when an existing one misbehaves. Live in all six languages.
← Back to Live Updates · 📚 Monthly archive