Benchgen

SWE-bench Pro — Results

RankModelScore
1flint89.7
2fugu-ultra73.7
3ling-3-1-flash-101968.74
4ling-3-1-flash-101968.74
5qwen3-8-max67.7
6beam-1-0-501b-a23b66.1
7hy4-preview65.7
8nex-n2-5-max65.7
9ornith-1-5-397b65.1
10gpt-5-6-sol64.6
11claude-opus-4-764.3
12gpt-5-6-terra63.4
13claude-sonnet-563.2
14gpt-5-6-luna62.7
15qwen3-8-flash-next62.5
16glm-5-262.1
17qwen3-8-27b61.7
18ripple-65b-preview61.42
19nex-n2-5-pro61.2
20sage-a1-561.2
21dots3-note-preview61
22nanbeige4-3-33b60.3
23ornith-1-5-35b-a3b59.6
24laguna-s-2-159.4
25fugu59

SWE-bench Pro — Phases

1 phaseActive

Phase 1·

SWE-bench Pro

Jun 22, 2026
Current