Qwen Code Benchmark Run Marred by Infrastructure Failures on SWE-bench
DennisYu07 · ghdev · 2026-08-13
The Qwen Code team released the results of a non-production full SWE-bench Verified E2E validation run for the qwen3.7-plus model.
- Status: The run was marked as QUARANTINED, and the final score was not published.
- Failure Analysis: Out of 500 total cases, only 13 resolved successfully, 8 were unresolved, and 7 had execution errors. A massive 472 cases failed due to infrastructure failures.
- Versions: The test utilized Qwen Code version v0.21.11, against the Benchmark-Qwen-Ref v0.21.11.
More from coding & agent
- Dispatch launches newsletter infrastructure built for AI agents, one API call to publish — RichardsonDx · 2026-10-16
- Hyperbrowser launches macOS sandboxes giving agents desktop, apps and Terminal control — madhavsinghal_ · 2026-10-16
- Creator shares the agent-powered system behind millions of monthly impressions — VibeMarketer_ · 2026-10-16
- Devs say LLM coding mistakes are a different class: overcomplication, inferred limits, worse at higher reasoning — Mu5_ · 2026-10-16
- How do you debug bad agent runs when you already have tracing? — Sensitive-Parsnip-12 · 2026-10-16
- Where does pre-task context live for agents? Devs debate workflows beyond handoff — shishen2026 · 2026-10-16