← All comparisons
Llama 3.1 Tulu3 405B vs Qwen3 235B A22B 2507 (Reasoning)
Allen Institute for AI vs Alibaba — side-by-side benchmark comparison
| Llama 3.1 Tulu3 405B | Qwen3 235B A22B 2507 (Reasoning) | |
|---|---|---|
| Intelligence Index | 14.1 | 29.5 |
| Coding Index | — | 23.2 |
| Math Index | — | 91.0 |
| Output speed (tok/s) | 0.0 | 62.5 |
| Blended price ($/1M) | $0.00 | $0.84 |
| Time to first token (s) | 0.00s | 1.21s |
| aime | 13.3% | 94.0% |
| aime 25 | — | 91.0% |
| artificial analysis coding index | — | 23.20 |
| artificial analysis intelligence index | 14.10 | 29.50 |
| artificial analysis math index | — | 91.00 |
| gpqa | 51.6% | 79.0% |
| hle | 3.5% | 15.0% |
| ifbench | — | 51.2% |
| lcr | — | 67.0% |
| livecodebench | 29.1% | 78.8% |
| math 500 | 77.8% | 98.4% |
| mmlu pro | 71.6% | 84.3% |
| scicode | 30.2% | 42.4% |
| tau2 | — | 53.2% |
| terminalbench hard | — | 13.6% |
Benchmark data from Artificial Analysis.