Insights from Benchmarking Frontier Language Models on Web App Code Generation

Benchmark Model Rank Results
code-generation-on-webapp1k-reactgpt-4o-2024-08-06#3pass@1: 0.885
code-generation-on-webapp1k-reactclaude-3.5-sonnet#4pass@1: 0.8808
code-generation-on-webapp1k-reactmistral-large-2#6pass@1: 0.7804
code-generation-on-webapp1k-reactdeepseek-coder-v2-instruct#7pass@1: 0.7002
code-generation-on-webapp1k-reactllama-v3p1-405b-instruct#8pass@1: 0.302