vercel/ai @ai-sdk/prodia@2.0.0-beta.5

on GitHub

Patch Changes

• e2bdcd6: feat(provider/prodia): Add LanguageModel and VideoModel support to the Prodia provider.

  • LanguageModel: Supports Nano Banana (inference.nano-banana.img2img.v2) for img2img generation with text+image output via multipart form-data requests. Implements both doGenerate and doStream.
  • VideoModel: Supports Wan 2.2 Lightning for text-to-video (inference.wan2-2.lightning.txt2vid.v0) and image-to-video (inference.wan2-2.lightning.img2vid.v0) generation.
  • Extract shared multipart parsing and error handling infrastructure into prodia-api.ts.