Intern-S1: benchmark results
Shanghai AI Lab's open scientific multimodal reasoning model on a Qwen3 235B MoE base. Provider: Shanghai AI Lab. Released 2025-07-27. Access: Open.
Unified ELO 1594 ± 1, rank #279 of 1392 rated models, from 15 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| SGI-Bench Deep Research | 15.74 | Deep Research Score | 94.4 |
| MMMU Benchmark | 77.7 | Validation Score | 91.5 |
| SGI-Bench Wet Experiment | 29.02 | Wet Experiment Score | 61.1 |
| SciEval - Life Sciences | 53.7 | Life Sciences (%) | 61.1 |
| SciEval - Materials Science | 81.69 | Materials Science (%) | 61.1 |
| SciEval - Earth Sciences | 68.52 | Earth Sciences (%) | 55.6 |
| SGI-Bench Dry Experiment | 28.79 | Dry Experiment Score | 44.4 |
| SciEval - Chemistry | 75.28 | Chemistry (%) | 44.4 |
| SciEval - Overall | 56.73 | Overall (%) | 44.4 |
| SGI-Bench Idea Generation | 38.09 | Idea Generation Score | 33.3 |
| SGI-Bench | 28.1 | SGI-Score | 27.8 |
| LiveSecBench | 31.46 | Overall Score (%) | 26.2 |
Interactive version: theaggregate.ai/model?slug=intern-s1 · How It Works · Data refreshed daily, snapshot 2026-09-05.