Gensyn Releases Auditable open-1b AI Model

Gensyn releases open-1b, an AI model that lets users inspect and independently verify its training, as detailed in a company press release. Gensyn says the release includes the model, its training recipe, and verifiable proof that allows third parties to audit any stage of training.

Gensyn distinguishes open-1b from open weight models, which provide model weights without the training process, and open recipe models, which include instructions but lack verifiable proof. The company says its infrastructure allows training results to be reproduced and verified across different hardware.

Co-founder and chief executive Ben Fielding described auditable AI as a way to inspect the data shaping a model, verify its training process and understand its conclusions without relying solely on a company's claims. Gensyn describes itself as a decentralized AI lab that builds forecasting systems that improve by learning from real world outcomes, and is backed by investors including a16z crypto, CoinFund and Galaxy Digital.