Intern-S1 places in the top three on CompassArena Multimodal.
Shanghai AI Lab Jul 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 | |
|---|---|---|---|---|
| GPQA Diamond | 77.3% | #134 of 302 | Lab model card | |
| AIME 2025 | 86% | #70 of 143 | Lab model card | |
| Matbench | 75.0 | #1 of 2 | Lab model card | |
| HiPhO | 52.2% | #21 of 36 | Official leaderboard | |
| SGI-Bench | 28.1% | #13 of 15 | Official leaderboard | |
| ChemBench | 83.4% | #1 of 1 | Lab model card | |
| PHYSICS | 44% | #1 of 1 | Lab model card |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| MMMU | 77.7% | #15 of 136 | Official leaderboard | |
| CompassArena Multimodal | 1,155 | #2 of 8 | Official leaderboard | |
| XLRS-Bench | 55% | #1 of 4 | Lab model card | |
| MathVista | 81.5% | #27 of 70 | Lab model card | |
| MSEarth | 65.7% | #1 of 2 | Lab model card | |
| MMStar | 74.9% | #21 of 36 | Lab model card | |
| MicroVQA | 63.9% | #2 of 3 | Lab model card | |
| MATH-Vision (MathVision) | 62.5% | #38 of 51 | Lab model card | |
| Scientists' First Exam (SFE) | 44.3% | #5 of 5 | Lab model card |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| LiveSecBench | 31.5 | #25 of 34 | Official leaderboard | |
| SMolInstruct | 51% | #2 of 2 | Lab model card | |
| ProteinLMBench | 63.1% | #1 of 1 | Lab model card |
The Self-Reported Index fits one score to the 8 capability results Shanghai AI Lab 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 135.1 to 140.5. How we compute it.
Where the benchmark's own leaderboard, or Epoch AI's run, also has this model. Gaps mostly come from different harnesses, effort levels and task subsets. Every pair.
The benchmarks the lab chose to headline, with the scores it gave.
Every result in the longer documents the lab published with it.