Chinese models
344 language models from Chinese developers, with prices at their developers and at hosts elsewhere. At 60% on GPQA Diamond, the cheapest Chinese model costs $0.054 per 1M tokens and the cheapest US model $0.028.
Prices on Chinese developers’ own platforms are in US dollars from the LiteLLM price list, except DeepSeek, whose official yuan prices (with peak and off-peak rates) are read from its own pricing page and converted at the European Central Bank rate. Hosts outside China are priced from OpenRouter.
Same quality, by developer country
- China
- United States
At home and abroad
The same model, bought from a Chinese host (the developer or a mainland cloud) and from the cheapest host elsewhere.
| Model | Chinese host | Elsewhere | Difference |
|---|---|---|---|
| MiMo-V2.6-Pro | $0.653Streamlake | $0.486DeepInfra, 5 hosts | ×1.3 cheaper abroad |
| Qwen3.8 Max (0902) | $3.00Alibaba Cloud | $2.48DeepInfra, 4 hosts | ×1.2 cheaper abroad |
| Kimi K3 | $6.00Moonshot AI | $2.28Sail Research, 25 hosts | ×2.6 cheaper abroad |
| Qwen3.7 Max | $2.21Alibaba Cloud | $1.88Novita, 4 hosts | ×1.2 cheaper abroad |
| GLM 5.3 | $1.08SiliconFlow | $0.409Wafer, 41 hosts | ×2.6 cheaper abroad |
| GLM 5.3 Flash | $0.069Streamlake | $0.054Openinference, 35 hosts | ×1.3 cheaper abroad |
| GLM 5.2 | $0.941Streamlake | $0.698DeepInfra, 27 hosts | ×1.3 cheaper abroad |
| DeepSeek V4.1 Flash | $0.257Streamlake | $0.113Decart, 35 hosts | ×2.3 cheaper abroad |
| GLM 5.1 | $1.48Streamlake | $1.51Chutes, 12 hosts | ×1.0 dearer abroad |
| Kimi K2.6 | $1.08Streamlake | $0.825Chutes, 21 hosts | ×1.3 cheaper abroad |
| Qwen3.7 Plus | $0.56Alibaba Cloud | $0.56Together | Same |
| DeepSeek V4 Pro 0423 | $1.19Streamlake | $1.05Relace, 16 hosts | ×1.1 cheaper abroad |
| GLM 5 | $0.93Streamlake | $0.93GMI Cloud, 7 hosts | Same |
| Kimi K2.5 | $1.00SiliconFlow | $0.90DeepInfra, 8 hosts | ×1.1 cheaper abroad |
| Qwen3.8 27B | $0.956Alibaba Cloud | $0.14Tenstorrent, 23 hosts | ×6.8 cheaper abroad |
| Step 5 Preview | $1.43StepFun | $1.43DeepInfra, 3 hosts | Same |
DeepSeek’s off-peak hours
- deepseek-v4-pro-0813 cached input50% off¥0.15 vs ¥0.3
- deepseek-v4-pro-0813 input50% off¥4.5 vs ¥9
- deepseek-v4-pro-0813 output50% off¥13.5 vs ¥27
- deepseek-v4.1-flash-1017 cached input50% off¥0.02 vs ¥0.04
- deepseek-v4.1-flash-1017 input50% off¥1 vs ¥2
- deepseek-v4.1-flash-1017 output50% off¥4 vs ¥8
Per 1M tokens in yuan, from DeepSeek’s Chinese pricing page. Off-peak hours are set by DeepSeek in Beijing time.
Models by developer
- Qwen139
- DeepSeek67
- Z.ai47
- Moonshot AI27
- MiniMax20
- Tencent Cloud7
- ByteDance7
- Xiaomi6
- ByteDance Seed6
- Baidu6
- 01 AI5
- StepFun3
Chinese language models, best rated first
| Model | Developer | Weights | Input | Output | Hosts | Arena | GPQA |
|---|---|---|---|---|---|---|---|
| MiMo-V2.6-Pro | Xiaomi | Open | $0.387 | $0.783 | 6 | 1491 | — |
| Qwen3.8 Max (0902) | Qwen | — | $1.65 | $4.95 | 5 | 1483 | 93% |
| Kimi K3 | Moonshot AI | Open | $0.238 | $5.50 | 28 | 1476 | 93% |
| Qwen3.7 Max | Qwen | Closed | $1.25 | $3.75 | 5 | 1475 | 91% |
| GLM 5.3 | Z.ai | Open | $0.079 | $1.40 | 46 | 1474 | 91% |
| GLM 5.3 Flash | Z.ai | Open | $0.0052 | $0.145 | 39 | 1471 | 90% |
| GLM 5.2 | Z.ai | Open | $0.034 | $1.44 | 33 | 1470 | 92% |
| MiMo-V2.5-Pro | Xiaomi | Open | $0.305 | $0.609 | 5 | 1465 | — |
| DeepSeek V4.1 Flash | DeepSeek | Open | $0.01 | $0.18 | 40 | 1464 | 90% |
| GLM 5.1 | Z.ai | Open | $0.966 | $3.04 | 17 | 1463 | 90% |
| Kimi K2.6 | Moonshot AI | Open | $0.45 | $1.95 | 25 | 1455 | 91% |
| MiMo-V2.6-Flash | Xiaomi | Open | $0.10 | $0.238 | 11 | 1455 | — |
| Qwen3.7 Plus | Qwen | Closed | $0.32 | $1.28 | 2 | 1455 | 88% |
| DeepSeek V4 Pro 0423 | DeepSeek | Open | $0.90 | $1.20 | 22 | 1451 | 91% |
| GLM 5 | Z.ai | Open | $0.60 | $1.92 | 11 | 1446 | 88% |
| Kimi K2.5 | Moonshot AI | Open | $0.45 | $2.25 | 10 | 1445 | — |
| Qwen3.8 27B | Qwen | Open | $0.02 | $0.50 | 24 | 1441 | — |
| Step 5 Preview | StepFun | Open | $1.00 | $2.70 | 4 | 1440 | — |
| GLM 4.6 | Z.ai | Open | $0.40 | $1.60 | 6 | 1440 | — |
| Qwen3-Max | Qwen | Closed | $1.20 | $6.00 | 2 | 1439 | 73% |
| Hy3 | Tencent Cloud | Open | $0.132 | $0.528 | 6 | 1438 | — |
| Qwen3.5 397B A17B | Qwen | Open | $0.39 | $2.34 | 13 | 1438 | 86% |
| GLM 5V Turbo | Z.ai | — | $1.20 | $4.00 | 2 | 1438 | — |
| Qwen3.6 Plus | Qwen | — | $0.325 | $1.95 | 2 | 1436 | 88% |
| GLM 4.7 | Z.ai | Open | $0.32 | $1.40 | 8 | 1435 | 83% |
| MiniMax M3 | MiniMax | Open | $0.224 | $0.88 | 18 | 1433 | 91% |
| DeepSeek V4 Flash 0423 | DeepSeek | Open | $0.05 | $0.14 | 27 | 1432 | — |
| GLM 4.5 | Z.ai | Open | $0.40 | $1.60 | 3 | 1430 | — |
| MiMo-V2.5 | Xiaomi | Open | $0.119 | $0.238 | 4 | 1428 | — |
| R1 0528 | DeepSeek | Open | $0.20 | $0.25 | 14 | 1428 | 76% |
| DeepSeek V3.2 | DeepSeek | Open | $0.209 | $0.31 | 18 | 1425 | — |
| Qwen3 VL 235B A22B Instruct | Qwen | Open | $0.20 | $0.88 | 7 | 1420 | — |
| deepseek-v3.1 | DeepSeek | Open | $0.25 | $0.95 | 9 | 1420 | — |
| Qwen3-235B-A22B-Instruct-2507 | Qwen | — | $0.09 | $0.55 | 7 | 1419 | — |
| DeepSeek V3.1 Terminus | DeepSeek | Open | $0.27 | $1.00 | 4 | 1419 | — |
| Qwen3.5-122B-A10B | Qwen | Open | $0.25 | $1.75 | 6 | 1417 | — |
| Qwen3 Next 80B A3B Thinking | Qwen | Open | $0.15 | $1.20 | 1 | 1417 | — |
| Qwen3 Next 80B A3B Instruct | Qwen | Open | $0.072 | $0.88 | 8 | 1417 | — |
| Qwen3 235B A22B Instruct 2507 | Qwen | Open | $0.088 | $0.35 | 7 | 1415 | 80% |
| Qwen3 235B A22B Thinking 2507 | Qwen | Open | $0.45 | $3.50 | 1 | 1415 | — |
Quality data: LMArena (CC BY 4.0), Epoch AI (CC BY 4.0). Prices: OpenRouter, LiteLLM (MIT), provider pages. List prices in US dollars before discounts and taxes; other currencies converted at the European Central Bank reference rate. The figures on these pages are free to reuse under CC BY 4.0 with a link to this page.