Group

For these benchmarks, what percentage of problems do you estimate the top-performing AI model or agent will be able to solve by December'25?

no comments0

Forecast Timeline

AI2 Reasoning Challenge

Epoch's FrontierMath

Toloka's U-MATH

Toloka's µ-MATH

Graduate-Level Google-Proof Q&A

Make a Prediction

Independent benchmarks

median

PDF

Authors:

ilia.zaharov

Opened:

Feb 12, 2025

Closes:

Nov 30, 2025

Scheduled resolution:

Dec 14, 2025

JetBrains

Artificial Intelligence

These researchers used NPR Sunday Puzzle questions to benchmark AI ‘reasoning’ models

TechCrunch•Feb 17, 2025

These are the hottest AI models, what they do, and how to use them

TechCrunch•Feb 17, 2025

The hottest AI models, what they do, and how to use them

TechCrunch•Feb 18, 2025

Learn more about Metaculus NewsMatch

What will be the best performance on FrontierMath by December 31st 2025?

65.7

What will be the best score on Cybench by December 31st 2025?

70.3

What will be the best normalized score achieved on the original 7 RE-Bench tasks by December 31st 2025?

1.25

Comments

? comments

Authors:

ilia.zaharov

Opened:

Feb 12, 2025

Closes:

Nov 30, 2025

Scheduled resolution:

Dec 14, 2025

JetBrains

Artificial Intelligence

These researchers used NPR Sunday Puzzle questions to benchmark AI ‘reasoning’ models

TechCrunch•Feb 17, 2025

These are the hottest AI models, what they do, and how to use them

TechCrunch•Feb 17, 2025

The hottest AI models, what they do, and how to use them

TechCrunch•Feb 18, 2025

Learn more about Metaculus NewsMatch

What will be the best performance on FrontierMath by December 31st 2025?

65.7

What will be the best score on Cybench by December 31st 2025?

70.3

What will be the best normalized score achieved on the original 7 RE-Bench tasks by December 31st 2025?

1.25