Benchmark Almanac

INTELLECT-3

INTELLECT-3.

Prime Intellect Released November 26th, 2025 Family INTELLECT The lab's model card Compare with others
11
Benchmarks with a score
0
Ranked first, out of five or more models
0
In the top three
Prime Intellect Nov 2025

INTELLECT-3

Open weights Flagship

138.2 #193 of 404
Self-Reported Index, from the lab's own numbers. No Epoch Capabilities Index score yet.
  • MMLU-Pro 81.9% #61 /167
  • LiveCodeBench v6 69.3% #57 /90
  • AIME 2025 88% #61 /143

Where it ranks

Its best score on each benchmark, and where that ranks among every model with a score there. We rank each model once, at its best setting.

Knowledge and language

BenchmarkScoreRankSource
MMLU-Pro 81.9% #61 of 167 Lab model card
LMArena (Elo) 1,356 #172 of 335 Official leaderboard
Humanity's Last Exam 14.6% #123 of 154 Lab model card
BullshitBench 12.7% #94 of 107 Official leaderboard
BullshitBench V2 7.00% #121 of 129 Official leaderboard

Math and science

BenchmarkScoreRankSource
MATH-500 98.1% #6 of 70 Lab model card
AIME 2024 90.8% #17 of 88 Lab model card
AIME 2025 88% #61 of 143 Lab model card
GPQA Diamond 74.4% #154 of 302 Lab model card

Coding

BenchmarkScoreRankSource
LiveCodeBench v6 69.3% #57 of 90 Lab model card

Agents and tools

BenchmarkScoreRankSource
SnakeBench 18 #149 of 188 Official leaderboard

Its lab's numbers put it at 138.2, number 193 of 404

The Self-Reported Index fits one score to the 7 capability results Prime Intellect reported for this model, on the scale of the Epoch Capabilities Index. Labs choose which benchmarks to show and how to run them, so read it as the lab's claim, not an independent measurement. The 90% interval runs from 135.7 to 140.8. How we compute it.

Its launch post

The benchmarks the lab chose to headline, with the scores it gave.

Its system card

Every result in the longer documents the lab published with it. 1 of them isn't in the launch post.