🤖 In a head-to-head comparison, Claude Sonnet 3.5 shows a 12% higher pass@1 rate than GPT-4o on our new RES-Q benchmark! Discover how these models stack up in real-world code-editing tasks. 🔗⬇️

As expected, Qurrent Radio (daily AI generated podcasts) is talking about the OpenAI drama. Skip to 6 min 10 seconds into today's technology podcast to listen!

I'm building AI agents that code. While working on completing an objective recently, one of the agents found it's own source code, and ran it. Meaning it recursively launched a new batch of agents inside of itself..... Maybe this is how the world ends.