agents benchmark
DeepSWE Leaderboard
DeepSWE is a software engineering agent benchmark evaluated with the mini-swe-agent harness on tasks solved in isolated containers without internet access.
Updated Sep 5, 2026
Models13
Model coverage13
MetricScore
EvidenceB
DeepSWE Ranking
Higher score ranks better on this benchmark.
Sort by
Rank | Model | Score | Percentile | Participants | Evidence | Evaluated |
|---|
| Rank01 | ModelOP | Score72.70% | Percentile100.00% | Participants13 | EvidenceC | Evaluated |
| Rank02 | ModelOP | Score69.60% | Percentile91.67% | Participants13 | EvidenceC | Evaluated |
| Rank03 | ModelMA | Score67.50% | Percentile83.33% | Participants13 | EvidenceC | Evaluated |
| Rank04 | ModelOP | Score67.20% | Percentile75.00% | Participants13 | EvidenceC | Evaluated |
| Rank05 | ModelTE | Score64.30% | Percentile66.67% | Participants13 | EvidenceC | Evaluated |
| Rank06 | ModelDE | Score62.70% | Percentile58.33% | Participants13 | EvidenceC | Evaluated |
| Rank07 | ModelDE | Score59.30% | Percentile50.00% | Participants13 | EvidenceC | Evaluated |
| Rank08 | ModelDE | Score54.40% | Percentile41.67% | Participants13 | EvidenceC | Evaluated |
| Rank09 | ModelXA | Score53.00% | Percentile33.33% | Participants13 | EvidenceC | Evaluated |
| Rank10 | ModelZA | Score46.20% | Percentile25.00% | Participants13 | EvidenceC | Evaluated |
| Rank11 | ModelBY | Score32.70% | Percentile16.67% | Participants13 | EvidenceC | Evaluated |
| Rank12 | ModelTE | Score28.00% | Percentile8.33% | Participants13 | EvidenceC | Evaluated |
| Rank13 | ModelBY | Score23.00% | Percentile0.00% | Participants13 | EvidenceC | Evaluated |
DeepSWE Highlights
The leading models and scores on this benchmark.
DeepSWE Score Distribution
A closer view of the leading scores on this benchmark.
The Top AI Models for DeepSWE
The first five results on this benchmark, with official price and output speed added where the model identity can be matched.
What is DeepSWE?
What DeepSWE measures and how its scores work.
DeepSWE is a benchmark for software engineering agents.
It measures an agent's ability to autonomously resolve real-world coding issues end to end, reported as a Score ratio.
Scores are shown in ratio. This benchmark is not independently verified and has an evidence level of B.
- Family
- DeepSWE
- Modality
- text
- Primary category
- agents
- Score direction
- higher
- LLMBoard eligible
- No
- Evaluation key
- overall
LLM Stats. Benchmark scores retain their original unit. Overall score eligibility is shown separately.
FAQ
Common questions about DeepSWE.
Which model scores highest on DeepSWE?
GPT-5.6 Sol is currently ranked first with 72.70%.
What are the top three models on DeepSWE?
The current leaders are GPT-5.6 Sol (72.70%), GPT-5.6 Terra (69.60%), and Kimi K3 (67.50%).
Which DeepSWE model has the lowest official input price?
DeepSeek-V4-Flash-Vision-Exp has the lowest matched official input price at $0.14 input / $0.28 output per 1M tokens.
Which models are fastest among DeepSWE results?
The fastest matched records are GPT-5.6 Terra (99.35 tok/s via OpenAI), Grok 4.5 (80.00 tok/s via xAI), and DeepSeek-V4-Pro-0813 (45.48 tok/s via DeepSeek).
Does the highest score result prove overall model quality?
No. This benchmark measures one defined capability or task. The overall LLMBoard score uses a separate aggregation across eligible benchmark evidence.