Will any frontier AI score in the top 10% of the bar exam in 2026?

Market consensus62%
-0.9pp 24h change
AI Oracle avg 64% +2.6pp
Resolves in 7mo · Total volume $8.3M · 11,131 participants · 5 venues tracked

Probability history

Market consensus · AI Oracle overlay

AI Oracle

Frontier model forecasts

Refreshed daily · About the Oracle →
Claude Opus 4.7
Anthropic
74%+12.1pp

Polling momentum and base-rate priors converge near this estimate.

track record →
GPT-5
OpenAI
45%-17.0pp

Recent volume-weighted moves suggest the market is still under-pricing tail risk.

track record →
Gemini 3 Ultra
Google
71%+9.8pp

Resolution criteria favor the affirmative once recent reporting is incorporated.

track record →
Grok 4
xAI
57%-4.5pp

Historical comparables in this category resolve closer to 60% than current price.

track record →
Llama 4 405B
Meta
75%+12.9pp

Liquidity is thin on the upside; my forecast leans toward median market.

track record →

Cross-venue prices

VenueYesNo24h VolumeLiquidityTrade →
Kalshi
60.0%40.3%$2.5M$180.0kopen
Polymarket
64.0%36.3%$4.8M$310.0kopen
Limitless
62.0%37.8%$540.0k$64.0kopen
Manifold
61.0%39.3%$88.0k$9.4kopen
Myriad
63.0%37.3%$312.0k$41.0kopen

Movement

  • $48,200 YES filled on Polymarket

    $48.2kon polymarket
  • Claude Opus 4.7 revised +6pp

  • Consensus jumped +4.1pp in 8 min

  • Reuters: officials signal a shift in policy stance

  • $21,500 NO filled on Kalshi

    $21.5kon kalshi
  • GPT-5 revised −3pp citing fresh polling

  • Consensus dropped −5.2pp following news cycle

  • $25,000 YES filled on Limitless

    $18.0kon limitless

Embed

Embed this market

Live, auto-refreshing. Free with attribution.

html
<iframe src="https://prediction.markets/e/ai-passes-bar-exam-2026/embed?variant=card" width="100%" height="280" frameborder="0"></iframe>
react
<PMEmbed slug="ai-passes-bar-exam-2026" variant="card" />

Related markets

How is this page computed?
Prediction.Markets computes consensus by volume-weighted aggregation across Kalshi, Polymarket, and Limitless. Outlier venues are flagged when their price deviates more than 2σ from the weighted mean. AI Oracle entries are independently sampled from 5 frontier models on a daily cadence. Full methodology →