Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Software Engineering Task Resolution on SWE-Bench Verified (with Token Usage)
Loading...
61.6
Success Rate (SR)
Terminus 2
36.848
43.274
49.7
56.126
Jun 11, 2026
Success Rate (SR)
Average Token Usage (M)
Updated 1mo ago
Evaluation Results
Method
Method
Links
Success Rate (SR)
Average Token Usage (M)
Terminus 2
Model=Qwen3.5-35B-A3B,...
2026.06
61.6
1.47
HarnessBridge
Model=Qwen3.5-35B-A3B,...
2026.06
60.2
1.13
mini-SWE-agent
Model=Qwen3.5-35B-A3B,...
2026.06
59.8
5.92
Meta-Harness
Model=Qwen3.5-35B-A3B,...
2026.06
59.2
1.92
Qwen-Coder
Model=Qwen3.5-35B-A3B,...
2026.06
58.8
3.86
OpenHands
Model=Qwen3.5-35B-A3B,...
2026.06
52.6
1.96
Meta-Harness
Model=GLM-4.7-Flash, H...
2026.06
49.6
2.43
Terminus-KIRA
Model=Qwen3.5-35B-A3B,...
2026.06
46
9.77
HarnessBridge
Model=GLM-4.7-Flash, H...
2026.06
46
1.48
mini-SWE-agent
Model=GLM-4.7-Flash, H...
2026.06
45.4
3.91
Terminus 2
Model=GLM-4.7-Flash, H...
2026.06
45.2
1.51
OpenHands
Model=GLM-4.7-Flash, H...
2026.06
42
3.69
Qwen-Coder
Model=GLM-4.7-Flash, H...
2026.06
42
4.34
Terminus-KIRA
Model=GLM-4.7-Flash, H...
2026.06
37.8
6.78
Feedback
Search any
task
Search any
task