A new analysis shows the top coding agents on swe-bench verified now solve the same 285 of 500…
a new analysis shows the top coding agents on swe-bench verified now solve the same 285 of 500 tasks and fail the same 51, with solution sets nested at 0.935 — the leaderboard can no longer order its own top entries. the remaining gap is scaffold, not model, with within-model ranges up to 29.8 points.
benchmark chasing has hit its ceiling.
Context
The arXiv paper Coding Agents Have Converged (2609.17394) audits 254 SWE-bench submissions across four splits without running models, with an analysis pool retrieved on 30 July 2026 (Verified: 134 submissions, 500 instances). Its abstract says the leading two Verified entries each resolve 396 of 500 instances, and the top ten share 285 successes and 51 failures, leaving 164 instances that distinguish their outcomes.
The 285 and 51 refer to the top ten submissions in a 30 July 2026 leaderboard snapshot, not coding agents generally, and not current performance. Model and scaffold associations are observational, not causal, and only 61 of 134 Verified submissions carry a single usable model tag, so the gap is scaffold, not model overstates the paper. The 0.935 nesting value and the 29.8-point range were not found in the text read, so they are unverified. It is a non-peer-reviewed audit of public submissions, with no reruns. Benchmark chasing has hit its ceiling is the author's opinion.
Watch next
- Peer review and a later snapshot.
Sources
Provenance
The note above is reproduced unedited from the original post, first published on Threads on 18 September 2026 at 15:09 IST. Sources are the papers and datasets the note draws on.
View the original post ↗Embed this note
More notes
The air is now being asked to keep its own ledger
the air is now being asked to keep its own ledger: ecmwf’s aifs compo becomes the first ai model to forecast atmospheric composition globally every three hours, cleanair simulates 365 days of pm2.5 over china in ten seconds, and a unified framework maps six pollutants at one kilometer across the whole country. the air now files its own composition report.
read the note →The current is now being asked to draw its own map
the current is now being asked to draw its own map: china’s langya 2.0 predicts six ocean phenomena including internal waves and mesoscale eddies, a deep net called wenhai resolves eddies globally with air sea flux formulas built in, and scripps infers surface currents from the way temperature patterns deform in satellite images. the ocean now files its own circulation report.
read the note →The soil is now being asked to report its own carbon
the soil is now being asked to report its own carbon: a nix color sensor paired with generative data augmentation predicts soil organic carbon without a lab, random forest drives 74 percent of soil health mapping studies, and sentinel 2 tracks five year carbon change across france and italy from 922 samples. the dirt now files its own carbon account.
read the note →