AI Weekly — 2026-09-04 to 2026-09-11 | Open-weight races collide with closed-frontier launches

Anthropic's Claude Fable 5.1 and OpenAI's GPT-6 Astra widen the web dev gap over open-source models

Anthropic's Claude Fable 5.1 (81.2% SWE-bench, 25% cheaper) and OpenAI's GPT-6 Astra (74.1% DeepSWE) launched within 48 hours, establishing ~1,800-point gap over open-source models. This accelerates premium AI tool adoption, forcing teams to reconsider build-versus-buy decisions as open-source trails frontier by four months.