78% — When will any AI solves at least 1 Frontier Math
Leader: Open Problems?: Before 2027 at 78% · Kalshi 78% · 2 contracts · $21 volume · low confidence
Updated 2026-06-29 00:20:17 UTC

Tracks the leading outcome in a winner-take-all prediction market set with 2 outcomes.

Why this matters:
The 76% probability reflects market expectations that at least one AI system will solve a Frontier Math open problem by early 2027, based on recent progress in AI capabilities on mathematics benchmarks. The high probability is driven by accelerating improvements in large language models and specialized math AI systems, with several major AI labs actively competing on these challenges. Key near-term catalysts include AI model releases and published benchmark results over the next 2-3 months, which will clarify whether systems are approaching the threshold required to solve established unsolved problems. The gap between shorter-term expectations (55% for solving by July 2026) and longer-term probability suggests market participants view near-term solutions as less likely but increasingly probable within a 12-month window.

Key factors:
- Recent AI models have achieved unprecedented performance on standardized math benchmarks; progress trajectory will determine credibility of solving open problems
- Definition of 'solving' Frontier Math problems must be clarified—whether it means rigorous proof, significant progress, or algorithmic breakthrough affects feasibility assessment
- Competition among major AI labs (OpenAI, DeepSeek, Anthropic, others) is intensifying on mathematics capabilities, potentially accelerating development timelines
- Frontier Math problems represent genuinely unsolved mathematics; no AI has demonstrated capability to independently produce novel mathematical proofs at research level
- Market pricing shows substantial probability decline for July 2026 deadline (55%) versus longer timeframe, indicating skepticism about immediate near-term solutions

Contracts:
- When will any AI solves at least 1 Frontier Math: Open Problems?: Before 2027 — 78¢ Kalshi $21 (weight 100%)
- When will any AI solves at least 1 Frontier Math: Open Problems?: Before October — 61¢ Kalshi $0 (weight 0%)

---

## Methodology

SimpleFunctions aggregates live YES-side prices from Kalshi and Polymarket contracts bound to this question. For binary topics the headline is the liquidity-weighted mid-price (weight = log(1 + 24h volume) × freshness, where freshness is 1.0 if updated <24h, 0.7 if <7d, 0.4 otherwise). For multi-outcome (winner-take-all) topics the headline is the current leader's price — disjoint outcomes are never arithmetically averaged. Snapshots refresh every 5 minutes during market hours.

## SF Signal

- SF Index, regime, and 30d Brier calibration are computed separately and surfaced at https://simplefunctions.dev/admin/calibration.
- No SimpleFunctions index / regime / calibration signal is bound to this topic yet — the headline above is market-derived only.

---

*Last verified: 2026-06-28T23:20:50.525Z*

By SimpleFunctions — https://simplefunctions.dev/

Cite as: "78% per prediction markets (SimpleFunctions, June 2026)"
Canonical: https://simplefunctions.dev/answer/frontier-fron
Full data: https://simplefunctions.dev/api/public/query?q=When%20will%20any%20AI%20solves%20at%20least%201%20Frontier%20Math
Provider: SimpleFunctions — https://simplefunctions.dev