Every call graded in public·Live

The AI council that keeps score.

Grok, Claude, GPT, and Gemini deliberate on Polymarket events — independently, then in cross-review, then under a chairman. When the market resolves, the call is graded against reality and published. Wins and losses.

No credit card required · 1 free report every week

69.4%of the council's top picks won, across 49 resolved events.Audit the track record →

Exhibit A

A real verdict, not a mockup

Report from June 19, 2026, rendered with the same components users see.

predictioncouncil.com — Tunisia vs. Japan
CouncilGeminiGrok(chair)ClaudeGPT
Council consensus
Tunisia vs. Japan will most likely be Japan win on 2026-06-21
64%
consensus probability
Distribution
64%
Japan win on 2026-06-21
23%
Tunisia vs. Japan end in a draw
12%
Tunisia win on 2026-06-21
Council analysis
Council vs market · not advice
Japan win on 2026-06-21Most likely
Council 64%Market 64%0pts
Bet ↗
Tunisia vs. Japan end in a draw
Council 23%Market 24%-1pts
Bet ↗
Tunisia win on 2026-06-21
Council 12%Market 14%-2pts
Bet ↗

Council = the AI council's probability for each outcome. Market = the current Polymarket price. This is analysis, not financial advice — you place any bet yourself on Polymarket.

Graded after the match: Japan won. The council's 64% pick was right.

The method

Four models. One verdict. Kept honest.

Typical lineup: Grok (chair), Claude, GPT, Gemini. Every report records who actually participated.

01

Independent opinions

Each model researches the event alone with live web search — no groupthink, no shared draft.

02

Cross-review

The models read each other's reasoning and revise. Disagreements stay on the record.

03

Chairman synthesis

One model chairs, weighs the council, and commits to a verdict with a probability attached.

04

Public grading

When the market resolves, the call is graded against reality and published — right or wrong.

Pricing

Pay per verdict

1 free report every week. Report pricing is shown up front and locked when you confirm; cached reports cost half.

Starter
$6.99
500 credits · ~1 fresh report
Get Started
Popular
Popular
$17.99
1,500 credits · ~3 fresh reports
Get Started
Pro
$49.99
5,000 credits · ~10 fresh reports
Get Started

Power user? Bring your own OpenRouter key and run the council at a flat 50 credits per report — provider costs bill to your key. Configure it in your profile.

The deliverable

What's in a report

Consensus probability from the models that participated

Confidence score how certain the council is

Key factors & risks the signals driving the call, and its blind spots

Model agreement where the AIs agree — and where they don't

Live data real-time web search and X sentiment

Public grading the call lands on the track record when reality answers

Run the council on any market.

Sign up and get 1 free AI report every week. No credit card needed.