NL2Repo leaderboard
NL2Repo
2 models tested · Updated 2026-04-07 · Verified sources only
GLM-5.1 leads at 42.7%
1
GLM-5.1
Zhipu AI · HuggingFace/zai-org · 2026-04-07
New SOTA on NL2Repo for open-weight models. Up 6.8pp from GLM-5 (35.9%).
42.7%
2
MiniMax M2.7
MiniMax · Blog/MiniMax · 2026-03-18
Strong repo-level code understanding. Open-weight, 229B params.
39.8%
benchmark.space canonical public ledger of AI capability