Category ranking · Chinese language
Best LLMs for Chinese
Chinese capability is not the same as translated English. This ranking reflects performance on Chinese-language understanding, generation, idioms and culturally specific tasks, drawing on Chinese benchmarks such as SuperCLUE and OpenCompass.
Scores are edited estimates from public Chinese benchmarks. Data as of 2026-11-20.
| # | Model | Chinese | Overall | Intelligence | Multimodal | Price (in/out) | License |
|---|---|---|---|---|---|---|---|
| 1 | Alibaba Cloud · CN | 98 | 88.8 | 88 | 90 | $2 / $6 | Proprietary |
| 2 | ByteDance · CN | 95 | 82.7 | 81 | 86 | $0.30 / $1.20 | Proprietary |
| 3 | Moonshot AI · CN | 92 | 89.8 | 91 | 80 | $2 / $10 | Open weights |
| 4 | Z.ai (Zhipu) · CN | 91 | 84.5 | 84 | 70 | $0.60 / $2.20 | Open weights |
| 5 | Google · US | 90 | 89.7 | 92 | 97 | $2 / $10 | Proprietary |
| 6 | Anthropic · US | 88 | 93.3 | 98 | 88 | $4 / $20 | Proprietary |
| 7 | DeepSeek · CN | 88 | 83.6 | 82 | 55 | $0.15 / $0.60 | Open source |
| 8 | OpenAI · US | 86 | 90.1 | 95 | 93 | $10 / $50 | Proprietary |
| 9 | OpenAI · US | 85 | 89.3 | 90 | 89 | $2 / $10 | Proprietary |
| 10 | Anthropic · US | 85 | 88.9 | 89 | 84 | $2 / $10 | Proprietary |
| 11 | xAI · US | 78 | 84.3 | 86 | 82 | $2 / $6 | Proprietary |
| 12 | Mistral AI · EU | 76 | 80.5 | 80 | 74 | $0.70 / $2.10 | Open weights |
| 13 | Meta · US | 74 | 78.5 | 77 | 82 | $0.20 / $0.60 | Open source |