← All comparisons
Llama 3.3 Instruct 70B vs Claude 3.5 Sonnet (Oct '24)
Meta vs Anthropic — side-by-side benchmark comparison
| Llama 3.3 Instruct 70B | Claude 3.5 Sonnet (Oct '24) | |
|---|---|---|
| Intelligence Index | 14.5 | 15.9 |
| Coding Index | 10.7 | 30.2 |
| Math Index | 7.7 | — |
| Output speed (tok/s) | 88.1 | 0.0 |
| Blended price ($/1M) | $0.62 | $6.56 |
| Time to first token (s) | 0.59s | 0.00s |
| aime | 30.0% | 15.7% |
| aime 25 | 7.7% | — |
| artificial analysis coding index | 10.70 | 30.20 |
| artificial analysis intelligence index | 14.50 | 15.90 |
| artificial analysis math index | 7.70 | — |
| gpqa | 49.8% | 59.9% |
| hle | 4.0% | 3.9% |
| ifbench | 47.1% | — |
| lcr | 15.0% | — |
| livecodebench | 28.8% | 38.1% |
| math 500 | 77.3% | 77.1% |
| mmlu pro | 71.3% | 77.2% |
| scicode | 26.0% | 36.6% |
| tau2 | 26.6% | — |
| terminalbench hard | 3.0% | — |
Benchmark data from Artificial Analysis.