Sharing some of our most recent eval results from #C07JVFEJFGX ! Blog: https://www.all-hands.dev/blog/evaluation-of-llms-as-coding-agents-on-swe-bench-at-30x-speed Tweet: https://twitter.com/allhands_ai/status/1842217720969846997 TLDR: Use claude-3.5-sonnet for high quality, and Deepseek v2.5 for way less $ (0.003$ per issue on average w/ prompt caching is 🤯 )