Claude 3 Opus beats GPT-4 on coding benchmarks
Posted by Tulsidas Goswami · 2024-03-06
Anthropic's Claude 3 Opus scores higher than GPT-4 on MMLU (86.8), HumanEval (84.9), and GSM8K (95.0). The competition is fierce and developers win. Multi-model strategies are no longer premature optimization.