Omni API now runs three production AIO model groups

Omni API now exposes three production AIO model groups from the same task API and Console flow:

• Omni Image AIO for text-to-image, single-image edit, and dual-image edit.
• Omni Video AIO for text-to-video, image-to-video, first/last-frame video, and video-to-video.
• Omni Multi AIO for image upscaling, audio transcription, voice-clone TTS, VLM chat, and Talk to Prompt.

The goal is simple: keep the model catalog small, operationally clear, and ready for real user work. Users should not have to choose between many half-connected model entries. They choose a model group, choose a task type, upload inputs when needed, and track everything in Console.

Console-first workflow

Task history, task detail pages, credits, billing, usage, payments, and API keys are being consolidated into one Console-style sidebar. The model playground is still available for quick trials, but production usage should feel consistent whether a user creates a task through the browser or through the API.

Credits and outputs

The AIO groups use the same task lifecycle: create, upload signed inputs, enqueue, run, store outputs, and review results. Text and JSON outputs can be stored directly with the task record; downloadable image, video, and audio outputs are stored through R2-backed object URLs.

This keeps the frontend simpler and gives support/admin workflows one place to inspect task state.