Estebankiwi@estebankiwi

Folgen

Composer 2.5 Artificial Analysis Index

Composer 2.5 has surged 15 points on the Artificial Analysis Coding Agent Index, moving from 48 to 63. The development merits closer attention since the model now sits in third place overall. It trails only Claude Code running on Opus 4.7 Max and Codex paired with GPT 5.5 xHigh. Composer 2.5 also surpasses Cursor operating on Opus 4.7 as well as Cursor using GPT 5.5. A proprietary Cursor model is beating frontier models inside their own testing environments while operating at a fraction of the cost. In contrast, Gemini CLI with Gemini 3.1 Pro remains at the bottom with a score of 43. Google, what are we doing? Composer 2.5 stands as the most underrated coding model at present.

3mo

Estebankiwi@estebankiwi

Folgen

Composer 2.5 Artificial Analysis Index

Composer 2.5 has surged 15 points on the Artificial Analysis Coding Agent Index, moving from 48 to 63. The development merits closer attention since the model now sits in third place overall. It trails only Claude Code running on Opus 4.7 Max and Codex paired with GPT 5.5 xHigh. Composer 2.5 also surpasses Cursor operating on Opus 4.7 as well as Cursor using GPT 5.5. A proprietary Cursor model is beating frontier models inside their own testing environments while operating at a fraction of the cost. In contrast, Gemini CLI with Gemini 3.1 Pro remains at the bottom with a score of 43. Google, what are we doing? Composer 2.5 stands as the most underrated coding model at present.

3mo