posted an update

GMI Cloud with GenBlaze and B2 Integration update

Seedance 2.5 image-to-video is the latest addition to our GMI Cloud integration. Routed through GenBlaze using seedance-2-5-260628, it supports first frame, optional last frames, adaptive framing, 720p output, and 5-, 10-, or 15-second scenes.

Model Matrix

Capability GMI Cloud models through GenBlaze Backblaze B2 role
Text and vision GPT 5.6 Sol · Gemini 3.1 Pro · Qwen 3.8 Max Publishes vision inputs as provider-readable HTTPS assets
Image generation GPT Image 2 · Seedream 5 Pro · Nano Banana 2 · Nano Banana Pro Stores generated images at durable public URLs
Image editing GPT Image 2 Edit · Nano Banana 2 Edit · Nano Banana Pro Edit · BRIA Eraser · BRIA GenFill Hosts source images and completed edits
Video generation Veo 3.1 and Fast · Seedance 1.5, 2.0 and 2.5 I2V · Kling 3 Pro, Turbo, 1.6 Pro and 2.1 variants · Hailuo 02 Pro · Happy Horse 1.1 I2V Publishes input frames and stores generated clips and final renders
Speech ElevenLabs Multilingual TTS · OpenAI TTS Stores generated audio assets

Model availability is credential-scoped. GenBlaze enables only exact routes verified for the connected GMI Cloud API key.

Local Media Delivery Fix

Slow or provider-unreachable local media URLs no longer block the agent pipeline. When GMI Cloud cannot download a managed local asset, GenBlaze:

  1. Reads the media through the internal gateway.
  2. Uploads it through GMI Cloud’s signed upload flow.
  3. Receives a stable, provider-hosted public URL.
  4. Retries the model request with the new URL.

This recovery path covers vision inputs, image edits, first and last video frames, and other media-based agent operations.

Integration Flow

Backblaze B2 media → GenBlaze gateway → GMI Cloud model → processed output → durable B2 asset

  • GenBlaze handles routing, request translation, media recovery, submission, and polling.
  • Backblaze B2 provides public media delivery and persistent output storage.
  • Completed assets are copied from temporary provider URLs into B2.
  • Setup validates the B2 endpoint, bucket access, and upload permissions.
  • Master keys use the Native B2 API; application keys use the S3-compatible API.

You can checkout the latest code here

Log in or sign up for Devpost to join the conversation.