Which company has the best Math AI model end of May?

Platform comparison

Active sub-markets

Market context

The Chatbot Arena Math Leaderboard ranks large language models by their performance on mathematical reasoning tasks, with settlement determined by whichever company's model holds the highest position on 31 May 2026. The leaderboard uses Elo-style rating derived from head-to-head comparisons, making it a live, continuously updated ranking rather than a static benchmark. Current crowd probability of 11% suggests either high fragmentation among potential winners or confidence in a specific incumbent's dominance through the settlement window.

Historical precedent from prior Chatbot Arena markets shows that leaderboard leadership has shifted significantly between OpenAI, Anthropic, and emerging challengers within six-month windows. In late 2024 and early 2025, mathematical reasoning became a focal point for model differentiation, with Claude 3.5 Sonnet and GPT-4o competing closely on symbolic manipulation and proof-based tasks. The relatively low probability implies traders expect either OpenAI or Anthropic to maintain a clear lead, or that fragmentation across multiple strong contenders makes any single company's victory unlikely.

Traders should monitor scheduled model releases and capability announcements from major labs through May 2026, particularly any improvements to reasoning chains or mathematical tokenisation. Recent reports indicate both OpenAI and Anthropic are investing heavily in reasoning-focused training; any public benchmarks released before settlement could shift leaderboard positions materially. The Arena's methodology—based on user preference voting rather than fixed test sets—means that perceived improvements in mathematical clarity or correctness can propagate quickly into ranking changes, making announcement timing a critical variable for position shifts in the final weeks before settlement.

Methodology

We track Which company has the best Math AI model end of May? on the five venues with material liquidity for prediction markets. Live odds come from the Polymarket Polygon order book — the only source that ships real-time data under an open licence. For Kalshi, Betfair and Manifold we list platform attributes (fee, KYC, settlement, payment) instead of fabricated odds, because their APIs use non-comparable contract definitions.

Resolution & payout

At resolution the UMA oracle takes over: a proposer posts the outcome with a bond, any token holder can dispute within two hours. Without dispute the result is accepted and the smart contract distributes USDC instantly.

On Kalshi (CFTC-regulated) resolution runs through their in-house clearing engine in USD. Betfair Exchange settles after match end in the account's local currency. Manifold pays no cash — only its in-platform "mana" currency.

FAQ

Where can I trade this market with the lowest fees?

On Polymarket Tax UK, which mirrors the Polymarket order book at 0% fees. Kalshi charges up to 7% per trade; Betfair Exchange takes 2-5% commission on net winnings.

Is this market available outside the US?

Polymarket Tax UK is available in most jurisdictions where Polymarket isn't directly accessible. Polymarket itself is geo-blocked in the US/UK/EU. Always check local regulations.

How does resolution work?

Through the UMA Optimistic Oracle on Polygon: a proposer submits the outcome, a two-hour challenge window opens, and USDC payouts settle automatically once the result is final.

What's the difference between YES and NO shares?

A YES share pays $1.00 if the event happens, $0 otherwise. A NO share pays $1.00 if the event doesn't happen. The market price between 0¢ and 100¢ is the implied probability.

Do I need to KYC for this market?

Not under $1,500 of lifetime trading volume. Above that threshold, Polymarket Tax UK triggers a quick verification flow that finishes in minutes.