# Repository-level issue resolution on SWE-Bench Verified (held-out (sample of 150)) Source: [Repository-level issue resolution on SWE-Bench Verified (held-out (sample of 150))]() Collection index: [llms.txt]() Data updated: 2026-05-26T11:01:55.280Z Evaluation of model performance on repository-level software engineering issue resolution using the held-out 150-instance sample of SWE-Bench Verified, measured by task pass rate and average reasoning steps. A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. | Scope | Recorded value | | --- | --- | | Task | Repository-level issue resolution | | Dataset (including any stated split) | SWE-Bench Verified (held-out (sample of 150)) | | Metric | Task Pass Rate | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 12 of 12 | | Coverage | All results currently returned for this benchmark | Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol. ## Metrics - [Task Pass Rate](): Higher is better; unit: PERCENT. - [Average Reasoning Steps](): Lower is better; unit: SCALAR. ## Recorded results | Method | Task Pass Rate | Result date | Report paper | | --- | --- | --- | --- | | CODESKILL | 66 | Not recorded | 6a152968dbd7150506f7e373 | | Prompt Skill Mgmt. | 64.67 | Not recorded | 6a152968dbd7150506f7e373 | | SFT-CODESKILL | 64 | Not recorded | 6a152968dbd7150506f7e373 | | Subtask Memory | 61.33 | Not recorded | 6a152968dbd7150506f7e373 | | Prompt Skill Mgmt. | 58.67 | Not recorded | 6a152968dbd7150506f7e373 | | No-skill baseline | 57.33 | Not recorded | 6a152968dbd7150506f7e373 | | Prompt Skill Mgmt. | 56.67 | Not recorded | 6a152968dbd7150506f7e373 | | CODESKILL | 56 | Not recorded | 6a152968dbd7150506f7e373 | | SFT-CODESKILL | 54.67 | Not recorded | 6a152968dbd7150506f7e373 | | Subtask Memory | 52.67 | Not recorded | 6a152968dbd7150506f7e373 | | Prompt Skill Mgmt. | 50.67 | Not recorded | 6a152968dbd7150506f7e373 | | No-skill baseline | 46.67 | Not recorded | 6a152968dbd7150506f7e373 | ### CODESKILL Task Pass Rate: 66. Result ID: 6a157da1fbc373592c0dc414. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Skill Backbone":"Qwen3.5-4B","Frozen coding policy":"Qwen3.5-35B-A3B"}. This record does not assert independent reproduction. ### Prompt Skill Mgmt. Task Pass Rate: 64.67. Result ID: 6a157da1fbc373592c0dc412. [Reporting paper]() Recorded evaluation configuration: {"Skill Backbone":"GPT-5.4-mini","Frozen coding policy":"Qwen3.5-35B-A3B"}. This record does not assert independent reproduction. ### SFT-CODESKILL Task Pass Rate: 64. Result ID: 6a157da1fbc373592c0dc413. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Skill Backbone":"Qwen3.5-4B","Frozen coding policy":"Qwen3.5-35B-A3B"}. This record does not assert independent reproduction. ### Subtask Memory Task Pass Rate: 61.33. Result ID: 6a157da1fbc373592c0dc410. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Skill Backbone":"GPT-5.4-mini","Frozen coding policy":"Qwen3.5-35B-A3B"}. This record does not assert independent reproduction. ### Prompt Skill Mgmt. Task Pass Rate: 58.67. Result ID: 6a157da1fbc373592c0dc411. [Reporting paper]() Recorded evaluation configuration: {"Skill Backbone":"Qwen3.5-4B","Frozen coding policy":"Qwen3.5-35B-A3B"}. This record does not assert independent reproduction. ### No-skill baseline Task Pass Rate: 57.33. Result ID: 6a157da1fbc373592c0dc40f. [Reporting paper]() Recorded evaluation configuration: {"Skill Backbone":"-","Frozen coding policy":"Qwen3.5-35B-A3B"}. This record does not assert independent reproduction. ### Prompt Skill Mgmt. Task Pass Rate: 56.67. Result ID: 6a157da1fbc373592c0dc418. [Reporting paper]() Recorded evaluation configuration: {"Skill Backbone":"GPT-5.4-mini","Frozen coding policy":"GPT-5.4-mini"}. This record does not assert independent reproduction. ### CODESKILL Task Pass Rate: 56. Result ID: 6a157da1fbc373592c0dc41a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Skill Backbone":"Qwen3.5-4B","Frozen coding policy":"GPT-5.4-mini"}. This record does not assert independent reproduction. ### SFT-CODESKILL Task Pass Rate: 54.67. Result ID: 6a157da1fbc373592c0dc419. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Skill Backbone":"Qwen3.5-4B","Frozen coding policy":"GPT-5.4-mini"}. This record does not assert independent reproduction. ### Subtask Memory Task Pass Rate: 52.67. Result ID: 6a157da1fbc373592c0dc416. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Skill Backbone":"GPT-5.4-mini","Frozen coding policy":"GPT-5.4-mini"}. This record does not assert independent reproduction. ### Prompt Skill Mgmt. Task Pass Rate: 50.67. Result ID: 6a157da1fbc373592c0dc417. [Reporting paper]() Recorded evaluation configuration: {"Skill Backbone":"Qwen3.5-4B","Frozen coding policy":"GPT-5.4-mini"}. This record does not assert independent reproduction. ### No-skill baseline Task Pass Rate: 46.67. Result ID: 6a157da1fbc373592c0dc415. [Reporting paper]() Recorded evaluation configuration: {"Skill Backbone":"-","Frozen coding policy":"GPT-5.4-mini"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.