SWE-bench co-author shares how the benchmark went from lukewarm launch to must-watch
OfirPress · x · 2026-11-14
In the first episode of Kaggle's new podcast Evaluated, SWE-bench co-author Ofir Press recounts how the benchmark went from a lukewarm launch to one of the most influential benchmarks in AI coding.
His advice for anyone building "the next SWE-bench": start with people — find teammates you can bounce ideas off of — then tackle problems so hard they seem impossible.
More from Companies & People
- OpenAI Hosts Content Workshop for Codex Physical Builds Community — IsaiahBallah · 2026-11-14
- Notion Ran ZeroEntropy's Reranker in Production Long Before the Acquisition — CShorten30 · 2026-11-14
- CoRL 2026 reportedly extended by a day, with Austin airport floated as new venue — notmahi · 2026-11-14
- Investor on CNBC names Rene Haas and Ian Buck as top picks to succeed Jensen Huang at Nvidia — firstadopter · 2026-11-14
- Bengio's LawZero to Open Berlin Office, Lobbying Europe for Safe-by-Design AI — Yoshua_Bengio · 2026-11-14
- 'Stand with Jensen' letter to Nvidia board crashes under flood of signatures — GavinSBaker · 2026-11-14