The results were validated on a leaderboard hosted by Scale Labs
Yes, mini-swe-agent is also great on our new agentic DrugDiscoveryBench: https://labs.scale.com/leaderboard/drugdiscoverybench Simple, but a really great harness! @OfirPress @KLieret
Our mini-swe-agent is becoming a standard harness for running benchmarks across industry and academia, because it's easy to run & extend while achieving similar or better performance than Claude Code and Codex on benchmarks.
Users are excited about Mini-SWE-Agent's strong results on the DrugDiscoveryBench leaderboard because it shows promising AI advances for drug discovery tasks.
No Digg Deeper questions have been answered for this story yet.
@TuXinming @KLieret Yay that's awesome!
I'm still surprised with all the tasks that people make mini-swe-agent do! This is on a new benchmark for drug discovery!
Yes, mini-swe-agent is also great on our new agentic DrugDiscoveryBench: https://labs.scale.com/leaderboard/drugdiscoverybench Simple, but a really great harness! @OfirPress @KLieret
