QwenClawBench

QwenClawBench - internal real-user-distribution Claw agent benchmark (open-sourcing soon); temp=0.6, 256K ctx.

Category

Agentic Coding

Max Score

100.0

Score Type

percent

Active

Yes

Model Scores

Scores for QwenClawBench

Model Score Date Verified
Qwen3.6-35B-A3B
98.36%
16.04.2026 Verified
Gemma4-31B-it
76.07%
16.04.2026 Unverified
Qwen3.5-35B-A3B
88.34%
16.04.2026 Unverified
Gemma4-26B-A4B
69.94%
16.04.2026 Unverified
Qwen3.6-27B
100.00%
22.04.2026 Verified
Qwen3.5-397B-A17B
96.73%
22.04.2026 Unverified
Claude 4.5 Opus
97.75%
22.04.2026 Unverified
Qwen3.5-27B
97.55%
16.04.2026 Unverified
MiniCPM5-2B
78.53%
— Verified
LFM2.5-2.6B
30.27%
— Verified
Qwen3.5-2B
28.02%
— Verified
Gemma4-E2B
20.45%
— Verified
granite-4.2-3B
65.24%
— Verified
Nemotron-3-Nano-4B
25.15%
— Verified
Gemma4-E4B
24.95%
— Verified
LFM2.5-8B-A1B
4.50
— Verified
Qwen3.5-4B
66.67%
— Verified