DeepSWE v1.1 编程榜：Gemini 4 Argon 77.9% 居首

BreezeOg @BreezeOg1原推发布于 2026.10.11

在 X 上打开原推

Grok 4.7 isn't in the top 3 for coding. On DeepSWE v1.1, these are the models above it and they are better for a reason The first one on the list is suprising: - Gemini 4 Argon: 77.9% - GPT-6.1 Sol: 75.2% - Claude Opus 5.5: 74.2% - Grok 4.7: 71.0% On Terminal-Bench 4.0, the gap gets wider: GPT-6.1 Sol scored 56% and Grok 4.7 scored 26%. But in the matter of cost, Grok 4.7 costs $2 / $6 per million tokens. Fable 5.1 costs $10 / $50. It's not the best coder overall. But if you want something cheap and efficient, It's the best coder for the price.