← All comparisons
Tri-21B-think Preview vs Qwen3.5 35B A3B (Reasoning)
Trillion Labs vs Alibaba — side-by-side benchmark comparison
| Tri-21B-think Preview | Qwen3.5 35B A3B (Reasoning) | |
|---|---|---|
| Intelligence Index | 20.0 | 37.1 |
| Coding Index | 7.4 | 30.3 |
| Math Index | — | — |
| Output speed (tok/s) | 0.0 | 135.9 |
| Blended price ($/1M) | $0.00 | $0.69 |
| Time to first token (s) | 0.00s | 1.23s |
| aime | — | — |
| aime 25 | — | — |
| artificial analysis coding index | 7.40 | 30.30 |
| artificial analysis intelligence index | 20.00 | 37.10 |
| artificial analysis math index | — | — |
| gpqa | 53.8% | 84.5% |
| hle | 5.7% | 19.7% |
| ifbench | 47.1% | 72.5% |
| lcr | 14.7% | 62.7% |
| livecodebench | — | — |
| math 500 | — | — |
| mmlu pro | — | — |
| scicode | 17.8% | 37.7% |
| tau2 | 93.3% | 89.2% |
| terminalbench hard | 2.3% | 26.5% |
Benchmark data from Artificial Analysis.