← All comparisons
Llama 3 Instruct 8B vs Claude 4 Sonnet (Reasoning)
Meta vs Anthropic — side-by-side benchmark comparison
| Llama 3 Instruct 8B | Claude 4 Sonnet (Reasoning) | |
|---|---|---|
| Intelligence Index | 6.4 | 38.7 |
| Coding Index | 4.0 | 34.1 |
| Math Index | — | 74.3 |
| Output speed (tok/s) | 83.3 | 55.5 |
| Blended price ($/1M) | $0.07 | $6.56 |
| Time to first token (s) | 0.47s | 8.92s |
| aime | 0.0% | 77.3% |
| aime 25 | — | 74.3% |
| artificial analysis coding index | 4.00 | 34.10 |
| artificial analysis intelligence index | 6.40 | 38.70 |
| artificial analysis math index | — | 74.30 |
| gpqa | 29.6% | 77.7% |
| hle | 5.1% | 9.6% |
| ifbench | 24.6% | 54.7% |
| lcr | 0.0% | 64.7% |
| livecodebench | 9.6% | 65.5% |
| math 500 | 49.9% | 99.1% |
| mmlu pro | 40.5% | 84.2% |
| scicode | 11.9% | 40.0% |
| tau2 | 0.0% | 64.6% |
| terminalbench hard | 0.0% | 31.1% |
Benchmark data from Artificial Analysis.