Key Takeaways
- • Gensyn launches Delphi, an on-chain prediction market platform enabling real-time trading on machine learning model performance, backed by a16z crypto, Eden Block, Protocol Labs, CoinFund, Canonical, and Maven 11.
- • Delphi creates price discovery for AI model evaluation by allowing users to buy and sell stakes in open-source models competing on live benchmarks, with continuous liquidity via a symmetrical logarithmic market scoring rule (LMSR) mechanism.
- • Platform is live on Gensyn testnet with the Middleweight General Reasoning Benchmark market operational; mainnet launch and transition to real economic value forthcoming over coming weeks and months.
- • Market mechanism eliminates traditional friction: fully on-chain pricing with zero slippage exits, no opaque order books or centralized matching engines, and fully auditable settlement when benchmarks conclude.
- • Future expansion roadmap includes user-created markets across additional domains, vault staking for liquidity providers earning market-neutral yields, and basket trading across model families.
Summary
Gensyn, a decentralized AI computing infrastructure platform backed by prominent crypto venture firms, has launched Delphi, an open-market platform for machine intelligence that functions as a prediction market for machine learning models. The platform allows participants to trade stakes in open-source AI models as they undergo live evaluation on standardized benchmarks, creating what the team describes as "the first live market signal of model performance."
The core innovation addresses a market inefficiency: AI model evaluation has historically lacked real-time price discovery mechanisms. Delphi solves this by enabling users to purchase positions in models they believe will outperform on specific benchmark tasks. As evaluation results are published and models' scores update, market prices adjust accordingly, allowing traders to enter, exit, or rotate positions with continuous liquidity. Winners receive rewards at settlement, creating what Gensyn characterizes as "a flywheel that drives development in open source AI."
The platform's technical architecture emphasizes market microstructure design. Delphi deploys a symmetrical logarithmic market scoring rule (LMSR)—a sophisticated continuous liquidity mechanism—fully on-chain. This engineering choice eliminates traditional friction points: there is no slippage when exiting positions, no opaque centralized order books, and no market maker intermediation. All trades and pricing remain transparent and auditable on-chain. The platform currently operates on the Gensyn testnet with test tokens; mainnet transition with real economic value is expected in the coming weeks and months.
The initial live market centers on the Middleweight General Reasoning Benchmark, with plans to expand across additional domains and longer-running markets covering a broader range of AI capabilities. Future features include vault staking functionality, allowing liquidity providers to earn market-neutral yields, and the ability for users to create custom markets and deploy their own models. The founding team includes Harry Grieve and Ben Fielding, both with backgrounds in AI research and quantitative finance.
The core innovation addresses a market inefficiency: AI model evaluation has historically lacked real-time price discovery mechanisms. Delphi solves this by enabling users to purchase positions in models they believe will outperform on specific benchmark tasks. As evaluation results are published and models' scores update, market prices adjust accordingly, allowing traders to enter, exit, or rotate positions with continuous liquidity. Winners receive rewards at settlement, creating what Gensyn characterizes as "a flywheel that drives development in open source AI."
The platform's technical architecture emphasizes market microstructure design. Delphi deploys a symmetrical logarithmic market scoring rule (LMSR)—a sophisticated continuous liquidity mechanism—fully on-chain. This engineering choice eliminates traditional friction points: there is no slippage when exiting positions, no opaque centralized order books, and no market maker intermediation. All trades and pricing remain transparent and auditable on-chain. The platform currently operates on the Gensyn testnet with test tokens; mainnet transition with real economic value is expected in the coming weeks and months.
The initial live market centers on the Middleweight General Reasoning Benchmark, with plans to expand across additional domains and longer-running markets covering a broader range of AI capabilities. Future features include vault staking functionality, allowing liquidity providers to earn market-neutral yields, and the ability for users to create custom markets and deploy their own models. The founding team includes Harry Grieve and Ben Fielding, both with backgrounds in AI research and quantitative finance.
Go deeper than the headline
Research every firm in this story, track the ones that matter and get the daily briefing — free for 7 days.