BULLET BLOG
LATEST POSTS
UPDATES / BENCHMARKS / RELEASE NOTES
BULLET BLOG
A running log of results and product updates.
Jev is now in Bullet
Jev decides what to inspect, edit, test, and when a task is actually done. The coding model focuses on writing code.
READ ARTICLE →95.8% on SWE-bench Verified
Full benchmark writeup covering setup, results, and repository level breakdown across 500 verified instances.
READ ARTICLE →