Benchmark
MMMLU
Multilingual benchmarks compare model performance across languages. Scores are percentages. Higher is better.
- Models and systems measured
- 49
- Scores recorded
- 163
- Sources
- 2
- IQ weight
- 2.48
Score distribution
- 79.8%
- Mean
- 81.3%
- Median
- 10.1%
- Std dev
- 68.3%
- 10th percentile
- 89.6%
- 90th percentile
- 38.4%
- Min
This benchmark has a weight of 2.48 for chat models, measured over 45 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.
Leaderboard
| # | Model or system | Score |
|---|---|---|
| 51 | Claude Sonnet 414 non-English languages · no extended thinking· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 85.4% |
| 52 | GPT OSS 120Bofficial model card Table 2 · Portuguese · reasoning high · translated-answer parser· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 85.3% |
| 53 | Qwen3.5 35B A3B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 85.2% |
| 54 | GPT-4.5 Preview· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 85.1% |
| 55 | GPT OSS 120Bofficial model card Table 2 · Italian · reasoning high · translated-answer parser· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 85% |
| 56 | GPT OSS 120Bofficial model card Table 2 · French · reasoning high · translated-answer parser· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 84.6% |
| 57 | GPT OSS 120Bofficial model card Table 2 · Spanish · reasoning medium · translated-answer parser· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 84.6% |
| 58 | Qwen3 Max· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 84.4% |
| 59 | GPT OSS 120Bofficial model card Table 2 · Indonesian · reasoning high · translated-answer parser· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 84.3% |
| 60 | GPT OSS 120Bhigh· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 83.8% |
| 61 | GPT OSS 120Bofficial model card Table 2 · Italian · reasoning medium · translated-answer parser· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 83.7% |
| 62 | GPT OSS 120Bofficial model card Table 2 · Chinese · reasoning high · translated-answer parser· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 83.6% |
| 63 | GPT OSS 120Bofficial model card Table 2 · Japanese · reasoning high · translated-answer parser· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 83.5% |
| 64 | Gemma 4 12B It· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 83.4% |
| 65 | GPT OSS 120Bofficial model card Table 2 · French · reasoning medium · translated-answer parser· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 83.3% |
| 66 | GPT OSS 120Bofficial model card Table 2 · Portuguese · reasoning medium · translated-answer parser· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 83.3% |
| 67 | Claude Haiku 4.5· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 83% |
| 68 | Claude Haiku 4.5launch-page setting· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 83% |
| 69 | GPT OSS 120Bofficial model card Table 2 · German · reasoning high · translated-answer parser· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 83% |
| 70 | GPT OSS 120Bofficial model card Table 2 · Korean · reasoning high · translated-answer parser· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 82.9% |
| 71 | GPT OSS 120Bofficial model card Table 2 · Indonesian · reasoning medium · translated-answer parser· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 82.8% |
| 72 | GPT OSS 120Bofficial model card Table 2 · Arabic · reasoning high · translated-answer parser· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 82.7% |
| 73 | GPT OSS 120Bofficial model card Table 2 · Hindi · reasoning high · translated-answer parser· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 82.2% |
| 74 | GPT OSS 120Bofficial model card Table 2 · Chinese · reasoning medium · translated-answer parser· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 82.1% |
| 75 | GPT OSS 120Bofficial model card Table 2 · Japanese · reasoning medium · translated-answer parser· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 82% |
| 76 | GPT OSS 120Bofficial model card Table 2 · German · reasoning medium · translated-answer parser· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 81.7% |
| 77 | Seed 2.0 Mini· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 81.6% |
| 78 | GPT 4o· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 81.4% |
| 79 | GPT OSS 120Bofficial model card Table 3 · reasoning high · 14-language average · pass@1· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 81.3% |
| 80 | GPT OSS 20bofficial model card Table 2 · Spanish · reasoning high · translated-answer parser· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 81.2% |
| 81 | Qwen3.5 9B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 81.2% |
| 82 | GPT OSS 120Bofficial model card Table 2 · Bengali · reasoning high · translated-answer parser· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 80.9% |
| 83 | GPT OSS 120Bofficial model card Table 2 · Korean · reasoning medium · translated-answer parser· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 80.9% |
| 84 | GPT OSS 120Bofficial model card Table 2 · Spanish · reasoning low · translated-answer parser· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 80.6% |
| 85 | GPT OSS 20bofficial model card Table 2 · Italian · reasoning high · translated-answer parser· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 80.5% |
| 86 | GPT OSS 20bofficial model card Table 2 · Portuguese · reasoning high · translated-answer parser· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 80.5% |
| 87 | GPT OSS 120Bofficial model card Table 2 · Arabic · reasoning medium · translated-answer parser· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 80.4% |
| 88 | GPT OSS 20bofficial model card Table 2 · French · reasoning high · translated-answer parser· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 80.2% |
| 89 | GPT OSS 120Bofficial model card Table 2 · Hindi · reasoning medium · translated-answer parser· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 80% |
| 90 | GPT OSS 120Bofficial model card Table 2 · Portuguese · reasoning low · translated-answer parser· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 80% |
| 91 | GPT OSS 20bofficial model card Table 2 · Spanish · reasoning medium · translated-answer parser· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 79.7% |
| 92 | GPT OSS 120Bofficial model card Table 2 · French · reasoning low · translated-answer parser· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 79.6% |
| 93 | GPT OSS 120Bofficial model card Table 2 · Italian · reasoning low · translated-answer parser· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 79.5% |
| 94 | GPT OSS 20bofficial model card Table 2 · Indonesian · reasoning high · translated-answer parser· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 79.5% |
| 95 | GPT OSS 20bofficial model card Table 2 · Chinese · reasoning high · translated-answer parser· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 79.4% |
| 96 | GPT OSS 120Bofficial model card Table 3 · reasoning medium · 14-language average · pass@1· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 79.3% |
| 97 | GPT OSS 20bofficial model card Table 2 · Portuguese · reasoning medium · translated-answer parser· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 79.2% |
| 98 | GPT OSS 20bofficial model card Table 2 · Italian · reasoning medium · translated-answer parser· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 79% |
| 99 | GPT OSS 20bofficial model card Table 2 · Hindi · reasoning high · translated-answer parser· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 78.8% |
| 100 | GPT OSS 20bofficial model card Table 2 · Japanese · reasoning high · translated-answer parser· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 78.8% |
Showing 51–100 of 159 results.
Not ranked
The source marks these results as descriptive, not capability scores, so we list them by name instead of ranking them.
| Model or system | Score |
|---|---|
| Qwen3.5-397B-A17B-BaseQwen3.5 release/model card · base-model table· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 85.8% |
| Qwen3.7-Plus-BaseQwen3.8-Flash-Next release/model card · base-model table· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 84.5% |
| Qwen3.8-27B-BaseQwen3.8-Flash-Next release/model card · base-model table· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 79.7% |
| Qwen3.8-Flash-Next-BaseQwen3.8-Flash-Next release/model card · base-model table· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 84.9% |
Data credits: BenchBoard by SYuan03 (github.com/SYuan03/benchboard). MIT covers its code and data organization; no results licence is inferred. Numeric relays retain the original stating-source links. · Data by LLM Stats