SWE-ABS: ~1 in 5 "Solved" SWE-Bench Patches Are Semantically Incorrect.

Shared entities / Same source domain / Shared topic / What happened next

SWE-bench Verified is leaking answers, and 60.8% of the "solved" issues prove it

Both cover Bench, SWE; reported by the same outlet (arxiv.org); overlapping topics (agent, benchmark, coding, leaderboard, swe-bench).

Shared entities / Shared topic / What happened next

Cognition built a coding benchmark that real maintainers grade, and the best model scores 13.4%

Both cover Bench, SWE; overlapping topics (agent, benchmark, coding, number, solved); picks up the Bench thread on 2026-06-09.

Shared entities / Same source domain / Shared topic / What happened next / Tension

Argus beats a direct coding agent 78% vs 59% on SWE-Bench Pro at 1.41x tokens, and gets cheaper as it matures.

Both cover Bench, SWE; reported by the same outlet (arxiv.org); overlapping topics (agent, coding).

Shared entities / Shared topic / What happened next / Tension

Kimi K2.6: A 1-Trillion-Parameter Open-Weights Model Just Beat GPT-5.4 on SWE-Bench Pro

Both cover Bench, SWE; overlapping topics (agent, benchmark, coding, swe-bench); picks up the Bench thread on 2026-04-21.

Shared entities / Same source domain / Shared topic / What happened next

MindForge's 27B model hits 49.51% where frontier models fail

Both cover Bench, SWE; reported by the same outlet (arxiv.org); overlapping topics (agent, benchmark, coding).

Tencent Leaderboard Splits Same Model's Score by 13.5 Points

Both cover Bench, SWE; reported by the same outlet (arxiv.org); overlapping topics (benchmark, leaderboard, number).

CompactionRL makes context compaction a trained policy, +7 on SWE-bench.

Both cover Bench, SWE; reported by the same outlet (arxiv.org); overlapping topics (agent, coding, swe-bench).

Coding Benchmarks Conflate Model, Retrieval, and Harness

Both cover Bench, SWE; reported by the same outlet (arxiv.org); overlapping topics (agent, benchmark, coding).