WEB VERSION

AI NEWS PAGE

September 10, 2026 · Updated Sep 10, 9:45 AM PT

BENCHMARKS:

DeepSWE v1.1 — 1T Gemini 3.8 Flash [HIGH] · 74% · $2.36/task 1T GPT 6 Astra [XHIGH] · 74% · $6.52/task 1T Claude Opus 5 [MAX] · 74% · $11.84/task 4 GPT 5.6 Sol [MAX] · 73% · $6.46/task 5 Claude Fable 5 [XHIGH] · 70% · $13.41/task (Sep 3)