← All comparisons
Llama 3.2 Instruct 90B (Vision) vs Claude 4.5 Haiku (Non-reasoning)
Meta vs Anthropic — side-by-side benchmark comparison
| Llama 3.2 Instruct 90B (Vision) | Claude 4.5 Haiku (Non-reasoning) | |
|---|---|---|
| Intelligence Index | 11.9 | 31.0 |
| Coding Index | — | 29.6 |
| Math Index | — | 39.0 |
| Output speed (tok/s) | 47.1 | 105.8 |
| Blended price ($/1M) | $1.38 | $2.19 |
| Time to first token (s) | 0.48s | 0.65s |
| aime | 5.0% | — |
| aime 25 | — | 39.0% |
| artificial analysis coding index | — | 29.60 |
| artificial analysis intelligence index | 11.90 | 31.00 |
| artificial analysis math index | — | 39.00 |
| gpqa | 43.2% | 64.6% |
| hle | 4.9% | 4.3% |
| ifbench | — | 42.0% |
| lcr | — | 43.7% |
| livecodebench | 21.4% | 51.1% |
| math 500 | 62.9% | — |
| mmlu pro | 67.1% | 80.0% |
| scicode | 24.0% | 34.4% |
| tau2 | — | 32.5% |
| terminalbench hard | — | 27.3% |
Benchmark data from Artificial Analysis.