Genblaze
An open-source Python SDK for orchestrating generative AI media pipelines across video, image and audio providers, with a built-in, verifiable provenance manifest for every output.
Genblaze is a Backblaze-built Python SDK for building and orchestrating generative-media workflows across video, image and audio. A unified Pipeline API spans providers such as OpenAI, Google, Runway, Luma, ElevenLabs and Stability Audio, plus models served through platforms like GMI Cloud and NVIDIA NIM, so a pipeline can switch providers without rewriting its orchestration logic.
Every run produces a canonical provenance manifest that can be embedded directly into media files such as MP4, PNG and MP3, and persisted to Backblaze B2 or any S3-compatible store. A verify function checks the manifest's hash and requires every output asset to declare a valid SHA-256 checksum; outputs become checksum-covered when a provider returns bytes directly or when object storage transfers them into durable storage, while URL-only outputs do not pass verification unless separately re-hashed.
Other features include fallback model chains that retry on a model error, provider-contract tests intended to catch upstream drift, and replayable runs, since a manifest captures the full provenance, including provider, model, prompt and parameters, so a run can be reconstructed from it later.