Coding Agent Index v2.0 scores drop across the board: 20 of 21 configs decline
ArtificialAnlys · x · 2026-10-23
Artificial Analysis notes that Coding Agent Index v2.0 scores are lower than v1.5 for nearly every configuration: across 21 configs measured on both versions, the mean fell from 54.8 to 47.4, with 20 of 21 declining.
The main driver is FrontierCode v1.1 replacing DeepSWE — the same configs average 45.1% on FrontierCode vs 65.6% on DeepSWE — with stricter SWE-Atlas-QnA-AA grading contributing the rest. Terminal-Bench 4.0 is unchanged; scores across the two versions aren't directly comparable.
More from coding & agent
- NVIDIA unveils OpenShell Reference Architecture to set boundaries for enterprise AI agents — NicoVerderosa · 2026-10-23
- Most 'unauthorized' AI agents in US bank accounts were let in by customers themselves, per Axios — nptacek · 2026-10-23
- Personal AI agents become a capital-intensive race as cloud execution environments reshape cost structures — sujingshen · 2026-10-23
- LLMs make code so cheap that recursive rewrites could make verified software tractable — jd_pressman · 2026-10-23
- Hugging Face hands OpenEnv RL environments project to Linux Foundation for neutral governance — lmoroney · 2026-10-23
- Theo writes the definitive explainer on what an AI coding "harness" actually is — pvncher · 2026-10-23