Coding Agent Leaderboard Updated
WenhuChen · x · 2026-07-09
Artificial Analysis has updated its Coding Agent Index, replacing the previous SWE-Bench Pro with Datacurve's DeepSWE. Following the update, GPT-5.5's Codex surpassed Claude Code with Opus 4.8 on the leaderboard, while the newly released Claude Fable 5 also debuted at the top within Claude Code.
The post further explains the reason for the swap: DeepSWE generates tasks from scratch rather than reusing public GitHub issues or PRs, making it much harder for models to simply "memorize the answers." In contrast, SWE-Bench Pro had become increasingly vulnerable to "cheating," as models could exploit the repository's commit history to solve tasks.
Related event: OpenAI Launches GPT-5.6 Series: Multi-Agent and Cost-Efficiency(119 posts)→
More from coding & agent
- Rowboat launches as an open-source, local-first AI coworker with memory — ycombinator · 2026-07-22
- Scoble says AI “loops” really means long-running multi-agent workspaces — Scobleizer · 2026-07-22
- Open-source runtime lets each repo define its own AI code reviewer — ibabufrik · 2026-07-22
- Indie Dev Asks: What's Actually Broken in Your AI Agent's Memory Today? — AcceptableTime7937 · 2026-07-22
- Fractal adds recursive agent loops for complex multi-step workflows — ryanpettry · 2026-07-22
- ACM essay says AI did not make programming easier, only differently difficult — tchalla · 2026-07-22