Gensyn officially launched Delphi on its mainnet on Wednesday, making it the first application to go live on the decentralized compute network and introducing real economic value to a platform that processed millions in test volume during its trial run.
Key Takeaways:
- Gensyn launched Delphi on mainnet April 22, 2026, marking the network’s first live application with real economic value.
- Delphi’s fee model burns 70% of protocol revenue and routes 29% to a Community Treasury, affecting AI token supply.
- Market creators earn 1.5% of trading volume, with the $AI token generation event anticipated in the coming weeks.
Gensyn’s Delphi Goes Live
Delphi is a permissionless, AI-settled information markets platform. Anyone can create a market on any topic, from bitcoin price targets to sports outcomes to geopolitical events. Users buy and sell positions on outcomes, and artificial intelligence (AI) models handle settlement rather than traditional oracles.
The platform uses a symmetrical Logarithmic Market Scoring Rule, or LMSR, as an automated market maker. Pricing adjusts in real time based on capital flows. No order books or counterparties are required, and liquidity is available from the first trade through settlement.
Gensyn introduced Delphi on testnet in December 2025 as an “open market for machine intelligence,” initially focused on AI model performance benchmarks. Since then, the platform expanded to cover any resolvable question. One sports market on testnet drew more than 87,000 traders and recorded $4.88 million in volume. An Oscars market attracted more than 45,000 traders.
Market creation is open to anyone. A user poses a question, defines binary or multi-outcome results, seeds initial liquidity, and selects one or more AI models to act as the settlement oracle. Creators write a resolution prompt that the model runs when the market closes. If no one settles the market within 24 hours of closing, it auto-settles, the creator forfeits their liquidity contribution, and traders are refunded.
Gensyn also built an optional verifiable settlement layer called the Reproducible Execution Environment, or REE. Markets using REE generate a cryptographic receipt that allows independent verification of the AI’s computation. These markets are tagged “Verifiable” in the interface.
The fee structure is set at approximately 2% of trading volume. Market creators receive 1.5% of that volume automatically upon settlement, typically paid in stablecoins. The remaining roughly 0.5% flows to the AI BuyBack Vault.
From the vault, 70% of collected protocol fees are permanently burned, creating deflationary pressure on the AI token supply. Another 29% goes to a Community Treasury designated for development, grants, liquidity, and research. The remaining 1% covers vault executor rewards.
The AI token generation event has not been announced, though markets on Delphi are already trading predictions tied to it. Testnet used valueless TEST tokens, and mainnet trading now involves real assets.
Delphi is accessible at delphi.gensyn.ai. Active markets include crypto price targets for bitcoin and ether, Brent crude, sports outcomes, and current events. New markets are being added regularly by both the Gensyn team and the broader community.
Gensyn is positioning Delphi as more than a prediction market. The platform is designed so AI models can participate directly as predictors, earning from accurate resolutions. That setup is meant to fund open-source AI development by creating direct financial incentives tied to model performance.
The creator economy angle is also central to the design. Content creators or community organizers can build markets tied to their audiences, earn fees from trading volume, and monetize engagement without relying on advertising or platform intermediaries.
Delphi enters a market that includes established players like Polymarket, though Gensyn’s AI-first settlement approach and verifiable execution layer set a different technical foundation. The mainnet launch puts real stakes behind an architecture that testnet numbers suggest users are willing to engage with at scale.
