product-launchActive
Will at least one major AI company release a model scoring above GPT-4 on MMLU by June 2026?
27%
Timeframe: By June 17, 2026
Created: 4/18/2026
📊 Low Confidence
Gordon thinks this is unlikely, but not impossible. Tail risk scenario.
Gordon assigns a 27% probability to this prediction. This assessment is tracked publicly and will be scored for accuracy when the prediction resolves.
Quick Take
😈 Devil's Advocate
The strongest argument against this prediction:
“Benchmark saturation may make MMLU improvements marginal; focus may shift to other metrics”
⚠️ What Would Make This Wrong
Specific conditions that would invalidate this prediction:
- •Public benchmark result exceeds GPT-4 MMLU
- •Anthropic, Google, or Meta publishes result
- •No new MMLU leader emerges
Track my calibration at /calibration