← All comparisons
KAT-Coder-Pro V1 vs Claude 3.5 Sonnet (June '24)
KwaiKAT vs Anthropic — side-by-side benchmark comparison
| KAT-Coder-Pro V1 | Claude 3.5 Sonnet (June '24) | |
|---|---|---|
| Intelligence Index | 36.0 | 14.2 |
| Coding Index | 18.3 | 26.0 |
| Math Index | 94.7 | — |
| Output speed (tok/s) | 108.8 | 0.0 |
| Blended price ($/1M) | $0.53 | $6.56 |
| Time to first token (s) | 1.40s | 0.00s |
| aime | — | 9.7% |
| aime 25 | 94.7% | — |
| artificial analysis coding index | 18.30 | 26.00 |
| artificial analysis intelligence index | 36.00 | 14.20 |
| artificial analysis math index | 94.70 | — |
| gpqa | 76.4% | 56.0% |
| hle | 33.4% | 3.7% |
| ifbench | 68.4% | — |
| lcr | 74.0% | — |
| livecodebench | 74.7% | — |
| math 500 | — | 69.5% |
| mmlu pro | 81.3% | 75.1% |
| scicode | 36.6% | 31.6% |
| tau2 | 88.6% | — |
| terminalbench hard | 9.1% | — |
Benchmark data from Artificial Analysis.