Super interesting work from Zoom and colleagues.
This is a AI post classified by Jev as AI coding agents (research), kept by the AI Radar because it carries real work, not commentary.
Super interesting work from Zoom and colleagues. If you maintain a hand-built coding harness, there are some great insights here. (bookmark it) They held the execution loop of a coding harness fixed and varied planning, the action space, and context management one at a time. They did across 176 matched settings, four models, SWE-Bench Verified and Terminal-Bench 2.1. Context management pays off more as the context window tightens. Most of its benefit comes from preventing overflow failures rather than from better reasoning. Staging rule-based elision before LLM summarization gave the be
Posted by DAIR.AI (132.4k followers) 20 h ago · 124 likes · 9.4k views · view the original post on X. Kept by the AI Radar as AI coding agents. Tools mentioned: academy.dair.ai.
More AI work like this
- i made X-Ray Anything with Codex. paste text, edit the labels, and Jev marks each… — @nicdunz
- A GROK ENGINEER JUST SHOWED HOW TO RUN A 24/7 REPO OPS DESK WITH 6 AI WORKERS WHILE YOU… — @gippp69
- i made Text DeepDream with Codex: random sentences evolve toward a concept. — @nicdunz
- AI searched random sentences and found this. — @nicdunz
- Just had an epic experience with Grok Bot/Cursor. — @mikepat711
- 90% confident that a ship would win against a village 😭 — @nicdunz
- a lot of people are asking me about this combined usage dashboard for codex (and other… — @argofowl
- i used Codex to make 2 random Wikipedia articles compete over questions like “who would… — @nicdunz
Every post is read and classified by Jev (TypeSafe): what it is, which market it belongs to, and whether the link is a real tool. 27.2k posts from 4.8k X accounts over the last 14 days, 1.2k tools, 19 markets. Collected every 5 minutes, fully re-ranked every hour — last update 2026-09-19 18:45 UTC. Full method.