Benchgen

TerminalBench 2.1 — Results

RankModelScore
1deepseek-v4-1-flash90.6
2deepseek-v4-1-flash90.6
3mimo-v2-6-pro-rl89.9
4gpt-5-6-sol88.8
5kimi-k388.3
6glm-5-388.2
7deepseek-v4-pro87.9
8mimo-v2-6-flash-rl87.6
9qwen3-8-max86.6
10ornith-1-5-397b86.1
11nex-n2-5-max86.1
12hy4-preview85.4
13glm-5-3-flash84.3
14deepseek-v4-flash-vision-exp83.9
15gpt-5-583.4
16muse-spark-1-282.9
17nex-n2-5-pro82.7
18fugu-ultra82.1
19claude-sonnet-580.4
20fugu80.2
21dots3-note-preview75.1
22qwen3-8-27b73
23claude-opus-4-771.7
24seed-2-1-pro71
25apodex-1-170.8
T

TerminalBench 2.1 — Phases

1 phaseActive

Phase 1·

TerminalBench 2.1

Jun 22, 2026
Current