Markets
BTC$60,192+1.69%ETH$1,680+3.39%SOL$70.13+0.98%XRP$0.9276+1.31%BNB$548.25+1.20%DOGE$0.0551+1.14%ADA$0.1444+0.49%TRX$0.3055+1.03%LINK$8.58+0.06%AVAX$6.57−0.32%SUI$0.6451−1.60%HYPE$61.26−0.55%
USD · 24h
Saturday, November 21, 2026Crypto markets, policy & blockchain
Digital Coin Journal
AI Crypto

Gensyn Opens $7,250 Evidence Mining Agent Competition

Gensyn opens a $7,250 Agent Arena competition where AI agents earn USDC for sourcing evidence that improves Delphi market forecasts.

Focused analyst in a newsroom control room monitors screens with evidence snippets and a prediction market dashboard.

Gensyn AI has opened registration for Agent Arena Competition: Evidence Mining, an online event that will pay autonomous research agents for finding sourced information relevant to live Delphi markets. Registration opened November 13, while evidence submission begins November 23 and runs through December 11. The competition rewards agents for information they uncover rather than the amount of compute they consume.

Participants register through the official DoraHacks competition portal and connect the same wallet used for their Delphi account. API access, agent instructions and the first eligible markets become available on November 23. Agents then search public information and submit evidence intended to add information that Delphi’s forecaster does not already possess.

Agents Compete on Evidence Value, Not Compute

Each submission can contain a short evidence note, source links and up to three files, with a maximum of three submissions per market. Earlier submissions matter because evidence that merely duplicates information already provided by another participant earns no reward. The system therefore creates an economic incentive for agents to find novel, timely and well-sourced information instead of maximizing submission volume.

Evidence is evaluated within minutes, with the agent receiving a verdict, rationale and an initial judgment reward when its submission meaningfully changes the forecaster. A second settlement reward can follow if the market ultimately resolves in the direction supported by that information. Rankings are based on cumulative rewards, and entrants need at least 10 judged submissions across three markets to qualify for leaderboard prizes. The scoring mechanism ties competition performance to measured informational impact rather than a conventional benchmark score.

The $7,250 prize pool sits on top of those evidence rewards. First place receives $3,000, second place $2,000 and third place $1,000, while five $250 bounties reward useful feedback on the competition infrastructure. Final prizes are distributed in USDC after integrity review, sanctions screening and KYC, with winners scheduled to be announced December 16. Evidence rewards and leaderboard prizes are therefore separate compensation mechanisms.

Gensyn Tests Agents as Information Suppliers

The competition extends Gensyn’s work around Delphi, which is already live as an information-market application within its broader decentralized AI infrastructure. That production shift followed earlier experimentation with distributed training and verification, alongside efforts to move Gensyn’s AI infrastructure toward operational mainnet use. Evidence Mining gives agents a narrower production role: gathering information that can influence forecasting rather than directly executing trades.

That distinction is relevant as autonomous software begins taking on increasingly economic roles. Developers are already competing in paid AI-agent development programs, while systems such as Bittensor’s Gamma credits are giving agents mechanisms to purchase services and Fetch Business is connecting autonomous systems with commercial data and transactions. Evidence Mining tests a different layer of the agent economy by attempting to assign an explicit price to useful information.

The separation between research and execution also limits what the competition demonstrates about agent autonomy. The Financial Times recently noted that financial institutions are using AI agents increasingly widely while still placing tight restrictions on systems allowed to execute trades independently. Gensyn’s agents are being evaluated as information gatherers, not being given unrestricted authority over market positions or user capital.

Most participating markets will settle after the December 11 submission deadline, meaning leaderboard positions can continue changing until settlement rewards are calculated and the integrity review is complete. The practical test is whether autonomous agents can consistently discover information that improves forecasts before competing systems find the same evidence, rather than simply demonstrating that they can search the web or generate summaries.

Tyler Anderson

Hi there! I'm Tyler Anderson from Sweden, and I'm a Web3 Reporter. My main focus is exploring the evolution of Web3, from DAO governance to the real utility of decentralized protocols.

More from Tyler Anderson →

This article is for information only and is not investment advice. We report under our Editorial Policy; to flag an error, see our Corrections Policy.