MGSM Thai
Math · 2022-10-06
Thai language split of the MGSM multilingual grade-school math benchmark.
Top models (higher is better)
| Model | Score |
|---|---|
| GPT-5.2 | 97.6 |
| Opus 4.5 | 96.4 |
| Gemini 3 Flash Preview | 96.4 |
| gpt-oss-120b | 96.4 |
| Grok 4 | 96.0 |
| Gemini 3 Pro Preview | 95.2 |
| GPT-5 | 95.2 |
| Claude 3.7 Sonnet | 94.8 |
| Gemini 2.5 Flash Preview (09-2025) | 94.8 |
| o3 | 94.8 |
| Qwen3-235B-A22B | 94.8 |
| Qwen3 Max Preview | 94.8 |
| Sonnet 4.5 | 94.4 |
| DeepSeek-R1 | 94.4 |
| GPT-5.1 | 94.4 |
| GPT-5 Mini | 94.4 |
| o3-mini | 94.4 |
| O4 Mini | 94.4 |
| Qwen3 Max (rolling alias) | 94.4 |
| Opus 4.1 | 94.0 |
| Opus 4 | 94.0 |
| Grok 4 Fast | 94.0 |
| O1 | 94.0 |
| Sonnet 4 | 93.6 |
| GLM-4.7 | 93.6 |
| Llama 3.3 70B Instruct | 93.6 |
| Haiku 4.5 | 93.2 |
| Gemini 2.0 Flash 001 | 93.2 |
| Gemini 2.5 Flash-Lite Preview (09-2025) | 93.2 |
| GPT-4.1 | 93.2 |
| Grok 3 | 93.2 |
| Grok 4.1 Fast | 93.2 |
| DeepSeek-V3-0324 | 92.8 |
| DeepSeek-V3.2 | 92.8 |
| GLM-4.5 | 92.8 |
| GPT-4o (2024-11-20) | 92.8 |
| Grok 3 Mini | 92.8 |
| Llama 4 Maverick Instruct | 92.8 |
| GLM 4.6 | 92.4 |
| GPT-4o (2024-08-06) | 92.4 |