← All comparisons
Llama 3.1 Instruct 70B vs Claude 4.1 Opus (Non-reasoning)
Meta vs Anthropic — side-by-side benchmark comparison
| Llama 3.1 Instruct 70B | Claude 4.1 Opus (Non-reasoning) | |
|---|---|---|
| Intelligence Index | 12.5 | 36.0 |
| Coding Index | 10.9 | — |
| Math Index | 4.0 | — |
| Output speed (tok/s) | 35.3 | 44.7 |
| Blended price ($/1M) | $0.56 | $32.81 |
| Time to first token (s) | 0.54s | 1.63s |
| aime | 17.3% | — |
| aime 25 | 4.0% | — |
| artificial analysis coding index | 10.90 | — |
| artificial analysis intelligence index | 12.50 | 36.00 |
| artificial analysis math index | 4.00 | — |
| gpqa | 40.9% | — |
| hle | 4.6% | — |
| ifbench | 34.4% | — |
| lcr | 6.3% | — |
| livecodebench | 23.2% | — |
| math 500 | 64.9% | — |
| mmlu pro | 67.6% | — |
| scicode | 26.7% | — |
| tau2 | 15.2% | — |
| terminalbench hard | 3.0% | — |
Benchmark data from Artificial Analysis.