Benchmarks
📊 Chinese AI Benchmark Database
Real, recorded benchmark scores for Chinese AI models — ranked per benchmark across coding, writing, math, vision, reasoning, context length, latency and API price.
Uncategorised
📊 SWE-bench Verified
↑ higher is better · %| 1 | DeepSeek-V4-Pro · DeepSeek | 80.6 % |
📊 Terminal-Bench Hard
↑ higher is better · %| 1 | Qwen3.7-Max · Alibaba | 50.8 % |