Qwen 3.7 Max (1m)
A competitive mid-table result with 47/151 tasks solved at least once and 25/151 solved in all three attempts; strongest around large Python/Django application repairs plus localized Go security-scanner changes. 这是一个有竞争力的中游结果：151 题中至少一次解出 47 题，三次都解出 25 题；强项主要落在大型 Python/Django 应用修复以及边界相对清楚的 Go 漏洞扫描器改动。
How to read this result可以这样读
- Qwen 3.7 Max (1m) is best read as moderately stable: rank #9, 47 reached tasks, 25 stable solves.Qwen 3.7 Max (1m) 更适合读成中等稳定型：排名 #9，触达 47 题，稳定解出 25 题。
- Best suite signal: Open Library · release 013 at 8/10 (80.0%).最强 suite 信号：Open Library · release 013，8/10（80.0%）。
- Weakest visible area: Flipt · release 005 at 0/10 (0.0%).最弱可见区域：Flipt feature flag 服务 · release 005，0/10（0.0%）。
- Qoder adds more workflow structure around the model, so its stable wins should be read as model-plus-shell behavior.Qoder 给模型外面加了更强的工作流结构，因此稳定胜利更适合读成 model-plus-shell 的组合效果。
Qwen 3.7 Max (1m) is a moderately stable row around the #9 slot. The useful reading is not just the 31.62 score, but the split between 47 reached tasks and 25 stable solves.
The closest family reference is Qwen 3.5 plus at rank #10. Compared with that row, this one is 0.23 points ahead, with 4 fewer reached tasks and 3 more stable solves.
The suite split is asymmetric: Open Library · release 013 at 8/10 (80.0%) supplies the main body of wins, vuls · release 012 at 4/4 (100.0%) supplies the clean spike, and Flipt · release 005 at 0/10 (0.0%) is where that pattern stops. Qoder adds more workflow structure around the model, so its stable wins should be read as model-plus-shell behavior.
The bars are a tooling story as much as a model story: Qoder helps on structured repair loops, but the weak suite still shows where orchestration cannot rescue the patch.
Look at psrp connection plugin accepts undocumented extras, causing ambiguous and inconsistent configuration. and [Bug]: Cache Middleware Causing Authorization Bypass and Performance Issues as shell-behavior examples. The difference is not only model knowledge; it is whether the workflow keeps the patch disciplined enough to pass.
The audit trims 25 solved attempts from Qwen 3.7 Max (1m) but still keeps 77% of the solved set, so the suite shape remains useful even where individual wins are debatable.
The available audit keeps 84 of 109 initial solved attempts. Read this as a robustness check, especially when the audit sample is smaller than 453 attempts.当前可用复核保留了 109 次初始成功中的 84 次。这更适合作为稳健性检查，特别是在复核样本小于 453 次尝试时。
For Qoder-style use, the interesting part is how the shell converts model guesses into patches. Compare Open Library · release 013 at 8/10 (80.0%) with Flipt · release 005 at 0/10 (0.0%) before attributing the result to the base model alone. The 109/453 attempt score should be read as model plus Qoder workflow, especially when comparing it with direct Qwen rows.
Supporting suite table
| Suite | Repo | Solved | Pass^3 | Rate |
|---|---|---|---|---|
release-zh-012-future-architect-vuls |
future-architect/vuls | 4/4 | 3 | 100.0% |
release-zh-013-internetarchive-openlibrary |
internetarchive/openlibrary | 8/10 | 4 | 80.0% |
release-zh-015-internetarchive-openlibrary |
internetarchive/openlibrary | 5/10 | 3 | 50.0% |
release-zh-007-flipt-io-flipt |
flipt-io/flipt | 4/10 | 2 | 40.0% |
release-zh-010-future-architect-vuls |
future-architect/vuls | 4/10 | 2 | 40.0% |
release-zh-011-future-architect-vuls |
future-architect/vuls | 4/10 | 1 | 40.0% |
release-zh-005-flipt-io-flipt |
flipt-io/flipt | 0/10 | 0 | 0.0% |
release-zh-018-qutebrowser-qutebrowser |
qutebrowser/qutebrowser | 0/9 | 0 | 0.0% |
release-zh-017-navidrome-navidrome |
navidrome/navidrome | 0/5 | 0 | 0.0% |
release-zh-008-flipt-io-flipt |
flipt-io/flipt | 1/10 | 0 | 10.0% |
Qwen 3.7 Max (1m) 是一个排名 #9 附近的中等稳定型结果。它的重点不只是 31.62 分，而是 47 道触达题和 25 道稳定题之间的差距。
最接近的同系参照是排名 #10 的 Qwen 3.5 plus。和它相比，这一行最终分高 0.23 分，触达题少 4 个，稳定题多 3 个。
suite 分布是不对称的：Open Library · release 013，8/10（80.0%）贡献主要胜利，vuls 漏洞扫描器 · release 012，4/4（100.0%）贡献最干净高点，而Flipt feature flag 服务 · release 005，0/10（0.0%）标出这种模式停止的地方。Qoder 给模型外面加了更强的工作流结构，因此稳定胜利更适合读成 model-plus-shell 的组合效果。
这些柱子既是模型故事，也是工具故事：Qoder 能帮助结构化修复循环，但弱 suite 仍说明哪些地方不是编排层能救回来的。
可以把 psrp connection plugin 接受未文档化 extras，导致配置含糊且不一致。 和 [Bug]: Cache Middleware Causing Authorization Bypass and Performance Issues 当成 shell 行为样本：差异不只是模型懂不懂，也在于工作流能否把补丁约束到可通过状态。
复核从 Qwen 3.7 Max (1m) 中剔除了 25 次成功，但仍保留 77% 的成功集合，因此即便个别胜利有争议，suite 形状仍然有参考价值。
The available audit keeps 84 of 109 initial solved attempts. Read this as a robustness check, especially when the audit sample is smaller than 453 attempts.当前可用复核保留了 109 次初始成功中的 84 次。这更适合作为稳健性检查，特别是在复核样本小于 453 次尝试时。
对 Qoder-style 使用来说，重点是 shell 如何把模型猜测压成补丁。在把结果完全归因到底座模型之前，应先对照Open Library · release 013，8/10（80.0%）和Flipt feature flag 服务 · release 005，0/10（0.0%）。109/453 的单次尝试成功数应读成模型加 Qoder 工作流的结果，尤其要和直接 Qwen 行对照。
支撑这个判断的 suite 表
| Suite | Repo | 解出 | Pass^3 | 通过率 |
|---|---|---|---|---|
release-zh-012-future-architect-vuls |
future-architect/vuls | 4/4 | 3 | 100.0% |
release-zh-013-internetarchive-openlibrary |
internetarchive/openlibrary | 8/10 | 4 | 80.0% |
release-zh-015-internetarchive-openlibrary |
internetarchive/openlibrary | 5/10 | 3 | 50.0% |
release-zh-007-flipt-io-flipt |
flipt-io/flipt | 4/10 | 2 | 40.0% |
release-zh-010-future-architect-vuls |
future-architect/vuls | 4/10 | 2 | 40.0% |
release-zh-011-future-architect-vuls |
future-architect/vuls | 4/10 | 1 | 40.0% |
release-zh-005-flipt-io-flipt |
flipt-io/flipt | 0/10 | 0 | 0.0% |
release-zh-018-qutebrowser-qutebrowser |
qutebrowser/qutebrowser | 0/9 | 0 | 0.0% |
release-zh-017-navidrome-navidrome |
navidrome/navidrome | 0/5 | 0 | 0.0% |
release-zh-008-flipt-io-flipt |
flipt-io/flipt | 1/10 | 0 | 10.0% |