DeepSWE blows up the AI coding leaderboard, crowns GPT-5.5(venturebeat.com)

4 pointsby ripvanwinkle2 ay önce1 comments

DeepSWE blows up the AI coding leaderboard, crowns GPT-5.5

https://venturebeat.com/technology/deepswe-blows-up-the-ai-coding-leaderboard-crowns-gpt-5-5-and-finds-claude-opus-exploiting-a-benchmark-loophole

1 comments

achow2 ay önce

My feeling seems to be getting validated. Offlate GPT 5.5 seems to be more precise and to the point, whereas Opus 4.7 is verbose, not in bad way, but still is kind of distracting. This aligns with the report finding where they say Opus 4.7 looks at 'surrounding enviornment' and bring in those aspect during task completion.