← All comparisons
Llama 3.2 Instruct 90B (Vision) vs Qwen3.6 27B (Reasoning)
Meta vs Alibaba — side-by-side benchmark comparison
| Llama 3.2 Instruct 90B (Vision) | Qwen3.6 27B (Reasoning) | |
|---|---|---|
| Intelligence Index | 11.9 | 45.8 |
| Coding Index | — | 36.5 |
| Math Index | — | — |
| Output speed (tok/s) | 47.1 | 57.4 |
| Blended price ($/1M) | $1.38 | $1.35 |
| Time to first token (s) | 0.48s | 1.46s |
| aime | 5.0% | — |
| aime 25 | — | — |
| artificial analysis coding index | — | 36.50 |
| artificial analysis intelligence index | 11.90 | 45.80 |
| artificial analysis math index | — | — |
| gpqa | 43.2% | 84.2% |
| hle | 4.9% | 21.6% |
| ifbench | — | 67.6% |
| lcr | — | 68.7% |
| livecodebench | 21.4% | — |
| math 500 | 62.9% | — |
| mmlu pro | 67.1% | — |
| scicode | 24.0% | 39.8% |
| tau2 | — | 94.2% |
| terminalbench hard | — | 34.8% |
Benchmark data from Artificial Analysis.