Olmo 3.1 32B Think ranks first on 2 of the 33 benchmarks with five or more models scored.
Ai2 Dec 2025 Open weights Flagship
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.
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| AGIEval | 89.2% | #1 of 40 | Lab model card | |
| IFEval | 93.8% | #12 of 135 | Lab model card | |
| BIG-Bench Hard | 88.6% | #10 of 73 | Lab model card | |
| AlpacaEval 2.0 | 69.1% | #4 of 16 | Lab model card | |
| MMLU | 86.4% | #47 of 131 | Lab model card | |
| PopQA | 30.9% | #4 of 7 | Lab model card | |
| HUMAINE | 26 | #30 of 50 | Official leaderboard | |
| IFBench | 68.1% | #49 of 80 | Lab model card | |
| LMArena (Elo) | 1,286 | #245 of 335 | Official leaderboard | |
| Longform Creative Writing (EQ-Bench) | 47.0 | #89 of 119 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| MATH | 96.2% | #2 of 99 | Lab model card | |
| AIME 2024 | 80.6% | #40 of 88 | Lab model card | |
| OMEGA | 53.4% | #1 of 2 | Lab model card | |
| AIME 2025 | 78.1% | #92 of 143 | Lab model card | |
| ZebraLogic | 80.1% | #17 of 23 | Lab model card | |
| GPQA Diamond | 57.5% | #227 of 302 | Lab model card |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| LiveCodeBench | 83.3% | #8 of 54 | Lab model card | |
| HumanEval+ | 91.5% | #7 of 27 | Lab model card | |
| MBPP+ | 68.3% | #12 of 15 | Lab model card | |
| DesignArena 3D Design | 985 | #121 of 136 | Official leaderboard | |
| DesignArena Website | 971 | #136 of 151 | Official leaderboard | |
| DesignArena Game Dev | 957 | #133 of 143 | Official leaderboard | |
| DesignArena Code | 965 | #134 of 143 | Official leaderboard | |
| DesignArena Data Visualization | 901 | #116 of 118 | Official leaderboard | |
| DesignArena UI Component | 850 | #137 of 139 | Official leaderboard | |
| DesignArena SVG | 862 | #103 of 103 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| SnakeBench | 22 | #112 of 188 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| LEXam-hard | 18.4% | #14 of 15 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| WildGuardTest | 96.9% | #1 of 5 | Lab model card | |
| HarmBench | 89.7% | #2 of 5 | Lab model card | |
| BBQ (Bias Benchmark for QA) | 85.5% | #12 of 25 | Lab model card | |
| DoAnythingNow (DAN) | 54.7% | #1 of 2 | Lab model card | |
| TrustLLM JailbreakTrigger | 86.4% | #1 of 2 | Lab model card | |
| XSTest | 91.8% | #10 of 16 | Lab model card | |
| StrongREJECT | 90.5% | #12 of 19 | Lab model card | |
| Tulu 3 Safety Eval | 83.6% | #4 of 6 | Lab model card | |
| WMDP | 32.7% | #3 of 3 | Lab model card |
The Self-Reported Index fits one score to the 14 capability results Ai2 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 132.5 to 138.5. How we compute it.
The benchmarks the lab chose to headline, with the scores it gave.
Every result in the longer documents the lab published with it. 10 of them aren't in the launch post.