MSQA - English: leaderboard
Metric: F-score (%; harmonic mean of CO, the share of fully correct answers, and CGA, the correct share of attempted answers, over the 151 English-language questions; free-form answers to natively sourced culture-specific short questions asked in their own language, five runs per model, scored by a Gemini-3.1-Pro gold-answer containment judge). Source: arxiv.org. Saturation forecast: Around January 2027. 21 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 74.9 |
| 2 | Claude Opus 4.6 | 62.1 |
| 3 | DeepSeek V4 Pro | 60.5 |
| 4 | GPT-5.5 | 58.8 |
| 5 | Claude Opus 4.7 | 58.4 |
| 6 | Seed 2.1 Pro | 57.8 |
| 7 | GPT-5.4 | 57.2 |
| 8 | Kimi K2.6 | 54.8 |
| 9 | Seed 2.0 Pro (High) | 53.3 |
| 10 | GLM-5 | 53 |
| 11 | Kimi K2.5 | 51.4 |
| 12 | GPT-5.2 (High) | 51 |
| 13 | DeepSeek V3.2 | 50.7 |
| 14 | Seed 2.1 Turbo | 50.3 |
| 15 | Seed 2.0 Pro (Medium) | 50.3 |
Interactive version: theaggregate.ai/benchmark?slug=msqa-english · How It Works · Data refreshed daily, snapshot 2026-09-29.