← All comparisons
Llama 3.2 Instruct 11B (Vision) vs Qwen3 VL 30B A3B (Reasoning)
Meta vs Alibaba — side-by-side benchmark comparison
| Llama 3.2 Instruct 11B (Vision) | Qwen3 VL 30B A3B (Reasoning) | |
|---|---|---|
| Intelligence Index | 8.7 | 19.7 |
| Coding Index | 4.2 | 13.1 |
| Math Index | 1.7 | 82.3 |
| Output speed (tok/s) | 77.0 | 123.6 |
| Blended price ($/1M) | $0.24 | $0.34 |
| Time to first token (s) | 0.49s | 1.09s |
| aime | 9.3% | — |
| aime 25 | 1.7% | 82.3% |
| artificial analysis coding index | 4.20 | 13.10 |
| artificial analysis intelligence index | 8.70 | 19.70 |
| artificial analysis math index | 1.70 | 82.30 |
| gpqa | 22.1% | 72.0% |
| hle | 5.2% | 8.7% |
| ifbench | 30.4% | 45.1% |
| lcr | 11.7% | 40.7% |
| livecodebench | 11.0% | 69.7% |
| math 500 | 51.6% | — |
| mmlu pro | 46.4% | 80.7% |
| scicode | 11.2% | 28.8% |
| tau2 | 14.6% | 19.9% |
| terminalbench hard | 0.8% | 5.3% |
Benchmark data from Artificial Analysis.