Benchmark-Qwen-Ref: v0.21.13
Full end-to-end validation after wheelhouse bootstrap-parent fix: Release -> DSW Harbor -> SWE-bench Verified (500) -> same Release publication -> Terminal-Bench 2.0 (89) -> same Release publication.
Qwen Code is pinned to published release v0.21.13.
SWE-bench Verified
- Status: QUARANTINED
- Dataset:
swe-bench/swe-bench-verified@2; 500/500 completed - Results: 183 resolved, 76 unresolved, 3 execution errors, 238 infrastructure failures
- Score: not published (non-scoreable or quarantined run)
- Score denominator: valid grader results (183 + 76 = 259)
- Qwen Code:
v0.21.13(d95901597430); modelqwen3.7-plus - Run:
pool-3c5b931833b94e08; dispatch workflow: https://github.com/QwenLM/qwen-code/actions/runs/32012707567 - Case trajectories:
swe-bench-verified-dsw-eas-full-20260817-r3-trajectories.tar.gz(262 cases)