← All comparisons
Llama 3.3 Instruct 70B vs Claude 4.1 Opus (Reasoning)
Meta vs Anthropic — side-by-side benchmark comparison
| Llama 3.3 Instruct 70B | Claude 4.1 Opus (Reasoning) | |
|---|---|---|
| Intelligence Index | 14.5 | 42.0 |
| Coding Index | 10.7 | 36.5 |
| Math Index | 7.7 | 80.3 |
| Output speed (tok/s) | 88.1 | 44.5 |
| Blended price ($/1M) | $0.62 | $32.81 |
| Time to first token (s) | 0.59s | 8.55s |
| aime | 30.0% | — |
| aime 25 | 7.7% | 80.3% |
| artificial analysis coding index | 10.70 | 36.50 |
| artificial analysis intelligence index | 14.50 | 42.00 |
| artificial analysis math index | 7.70 | 80.30 |
| gpqa | 49.8% | 80.9% |
| hle | 4.0% | 11.9% |
| ifbench | 47.1% | 55.4% |
| lcr | 15.0% | 66.3% |
| livecodebench | 28.8% | 65.4% |
| math 500 | 77.3% | — |
| mmlu pro | 71.3% | 88.0% |
| scicode | 26.0% | 40.9% |
| tau2 | 26.6% | 71.4% |
| terminalbench hard | 3.0% | 34.3% |
Benchmark data from Artificial Analysis.