Codesota · Computer Code · Program Repair · Defects4JTasks/Computer Code/Program Repair
Program Repair · benchmark dataset · 2014 · EN

Defects4J: A Database of Real Faults in Java Programs.

Standard program repair benchmark with 835 real bugs from 17 open-source Java projects. Each bug has a fix and triggering test suite. Primary metric is the number of correctly fixed bugs (plausible and correct patches).

Paper ↗Submit a result ↵
§ 01 · Leaderboard

Best published scores.

5 results indexed across 1 metric. Shaded row marks current SOTA; ties broken by submission date.

Primary
correct-patches · higher is better
correct-patches· primary
5 rows
#ModelOrgSubmittedPaper / codecorrect-patches
01SRepairSUTDApr 2024SRepair: Utilizing Multiple LLM Agents for Automated Pro…101
02Claude Opus 4APIAnthropicMar 2026official-model-card89
03GPT-4oAPIOpenAIApr 2024SRepair: Utilizing Multiple LLM Agents for Automated Pro…82
04ChatRepairFudan UniversityJan 2024ChatRepair: A Conversational Approach to Automated Progr…78
05AlphaRepairETH ZurichAug 2022Less Training, More Repairing Please: Revisiting Automat…23
Fig 2 · Rows sorted by score within each metric. Shaded row marks SOTA. Dates reflect model or paper release where available, otherwise the date Codesota accessed the source.
§ 03 · Progress

3 steps
of state of the art.

Each row below marks a model that broke the previous record on correct-patches. Intermediate submissions are kept in the leaderboard above; only SOTA-setting entries are re-listed here.

Higher scores win. Each subsequent entry improved upon the previous best.

SOTA line · correct-patches
  1. Aug 24, 2022AlphaRepairETH Zurich23
  2. Jan 3, 2024ChatRepairFudan University78
  3. Apr 18, 2024SRepairSUTD101
Fig 3 · SOTA-setting models only. 3 entries span Aug 2022 → Apr 2024.
§ 04 · Literature

3 papers
tied to this benchmark.

Every paper below corresponds to at least one row in the leaderboard above. Click through for the arXiv preprint and, when available, the reference implementation.

§ 06 · Contribute

Have a score that beats
this table?

Submit a checkpoint and a reproduction script. We will run it, publish the score, and — if it takes the top — annotate the step on the progress chart with your name.

Submit a result ↵Read submission guide
What a submission needs
  • 01A public checkpoint or API endpoint
  • 02A reproduction script with frozen commit + seed
  • 03Declared evaluation environment (Python, deps)
  • 04One row per metric declared by this dataset
  • 05A contact so we can follow up on discrepancies