jevIMAGE
Jev tested on CUASpeedrun computer-use benchmark and falls short
Someone ran Jev on cuaspeedrun.com for computer use: it is 1.5x slower than Opus-5.5-xHigh, 10x less accurate, and takes 5x more steps, putting it at GPT-4 level.
Pranjal Aggarwal ✈️ COLM'26
@PranjalAggarw16Archived 2026.10.03Originally posted 2026.10.03
Pranjal Aggarwal ✈️ COLM'26 @PranjalAggarw16Originally posted 2026.10.03
Thought Jev is fast? We evaluated Jev on https://cuaspeedrun.com for computer use agents and found: 1. 1.5x slower than Opus-5.5-xHigh 2. 10x lower accuracy 3. 5x more steps That puts it in the GPT-4 league of performance and Astra-xHigh on time What and Why?? 🤯 🧵
x.com/pranjalaggarw16/status/2106091065388261599Archived 2026.10.03 10:05