Benchmarks · knowledge
Global-MMLU-Lite
Global-MMLU-Lite: multilingual MMLU subset; accuracy.
33 models · 42 results · data as of 2026-10-29 · artificialanalysis.ai ↗
Leaderboard
GPT-5 high
All results
42 rows| # | Model | Effort | Global-MMLU-Lite |
|---|---|---|---|
| 1 | GPT-5 | high | 90.7% |
| 2 | Opus 4.5 | none | 90.6% |
| 3 | GPT-5.1 | high | 90.6% |
| 4 | GPT-5 | medium | 90.4% |
| 5 | Gemini 2.5 Pro | 90.3% | |
| 6 | Qwen 3.5 397B-A17B | 90.0% | |
| 7 | GPT-5.2 | medium | 89.8% |
| 8 | GPT-5.1 Codex | high | 89.7% |
| 9 | Grok 4 | 89.5% | |
| 10 | GPT-5.4 Mini | xhigh | 89.4% |
| 11 | Sonnet 4.6 | none | 89.2% |
| 12 | Qwen 3.5 122B-A10B | 89.1% | |
| 13 | GPT-5 Mini | high | 87.4% |
| 14 | GPT-5.2 | none | 86.8% |
| 15 | Qwen 3.5 27B | 86.8% | |
| 16 | Qwen 3.5 397B-A17B | none | 86.5% |
| 17 | Qwen 3.5 35B-A3B | 86.4% | |
| 18 | Qwen 3.5 122B-A10B | none | 85.9% |
| 19 | GPT-5.1 Codex Mini | high | 85.0% |
| 20 | GPT-5 | minimal | 84.4% |
| 21 | MiniMax M2.5 | 84.2% | |
| 22 | Kimi K2.5 | none | 84.0% |
| 23 | MiniMax M2.1 | 84.0% | |
| 24 | Qwen 3.5 27B | none | 83.8% |
| 25 | Qwen 3 Max | none | 83.3% |
| 26 | gpt-oss-120b | high | 82.8% |
| 27 | DeepSeek V3.2 | none | 82.7% |
| 28 | Llama 4 Maverick | none | 82.5% |
| 29 | Qwen 3.5 35B-A3B | none | 82.4% |
| 30 | GLM 5 | none | 81.9% |