Claude 4 models are meaningfully better on code benchmarks than o3 — the best reasoning model so far. And Anthropic is bundling web search, a Python execution sandbox and a files API 🤯 https://x.com/AnthropicAI/status/1925591523502301353