backblaze-labs/genblaze

What it solves

Genblaze is a pipeline SDK designed to orchestrate generative AI workflows across video, image, and audio. It addresses the problem of provider lock-in by providing a unified API to swap between different AI media providers (like OpenAI, Google, Runway, and Luma) without rewriting the orchestration logic. Additionally, it solves the issue of media authenticity by building provenance manifests by default, ensuring that generated assets are trackable and verifiable.

How it works

The SDK uses a Pipeline and Step API to chain generation operations. Each step specifies a provider, model, and prompt. The system integrates with various storage backends (primarily Backblaze B2 and other S3-compatible stores) to automatically handle asset transfer and manifest upload. Every run generates a canonical provenance manifest containing the provider, model, prompt, and a SHA-256 hash of the asset, which can be embedded directly into media files or stored as a sidecar.

Who it’s for

It is intended for developers building production-grade generative media pipelines who need to maintain a record of how assets were created, easily switch between AI models, and durable storage for their outputs.

Highlights

• Unified Provider API: Supports a wide range of providers including GMI Cloud, NVIDIA NIM, OpenAI, Google, and ElevenLabs.
• Provenance by Default: Generates hash-verified manifests for every run to ensure tamper-evidence and authenticity.
• S3-First Storage: Native integration with Backblaze B2 and other S3-compatible stores for durable, credential-free asset URLs.
• Replayable Runs: Captures full provenance so that any generation run can be reconstructed via the CLI.
• Fallback Chains: Supports fallback_models to automatically retry on model errors, increasing pipeline reliability.