Changelog

June 2026

1. 2026-06-08

   HiAPI docs and model integrations now use the Unified Async Task API: create tasks with POST /v1/tasks, read results with GET /v1/tasks/:id, and use callback.url for completion notifications.

May 2026

1. 2026-05-28

   Persistent output storage docs published (EN + ZH), covering how to enable it, billing, and limits.

2. 2026-05-12

   Added the HappyHorse 1.0 text-to-video model — 720p/1080p, 3–15s, multiple aspect ratios.

3. 2026-05-11

   Model detail pages now expose an Agent Skill install entry, so Codex, Claude Code and others can call a single model directly.

April 2026

1. 2026-04-30

   gpt-image-2 gained resolution tiers (1K / 2K / 4K); Skills guide shipped and MCP docs improved.

2. 2026-04-29

   MCP remote-access docs completed, with image-input upload and per-client configuration notes.

3. 2026-04-25

   The main pricing page model module was rebuilt as “table + accordion”; docs updated to match.

4. 2026-04-16

   Added Qwen Image 2.0 and Seedance 2.0 model docs, plus video model billing notes.