DeepSWE v1.1: How Fable and Kimi K2.7 Performed on the long‑horizon coding benchmark

DeepSWE v1.1 exposes real-world strengths and gaps in agentic coding. Here’s what the latest DeepSWE run shows for Anthropic's Fable and Moonshot's Kimi K2.7 — and what it means for engineering teams.