← All comparisons
Claude 4.5 Haiku (Reasoning) vs Llama 3.1 Instruct 8B
Anthropic vs Meta — side-by-side benchmark comparison
| Claude 4.5 Haiku (Reasoning) | Llama 3.1 Instruct 8B | |
|---|---|---|
| Intelligence Index | 37.1 | 11.8 |
| Coding Index | 32.6 | 4.9 |
| Math Index | 83.7 | 4.3 |
| Output speed (tok/s) | 142.2 | 212.7 |
| Blended price ($/1M) | $2.19 | $0.10 |
| Time to first token (s) | 10.48s | 0.45s |
| aime | — | 7.7% |
| aime 25 | 83.7% | 4.3% |
| artificial analysis coding index | 32.60 | 4.90 |
| artificial analysis intelligence index | 37.10 | 11.80 |
| artificial analysis math index | 83.70 | 4.30 |
| gpqa | 67.2% | 25.9% |
| hle | 9.7% | 5.1% |
| ifbench | 54.3% | 28.6% |
| lcr | 70.3% | 15.7% |
| livecodebench | 61.5% | 11.6% |
| math 500 | — | 51.9% |
| mmlu pro | 76.0% | 47.6% |
| scicode | 43.3% | 13.2% |
| tau2 | 54.7% | 16.4% |
| terminalbench hard | 27.3% | 0.8% |
Benchmark data from Artificial Analysis.