IFBench

IFBench - instruction following benchmark. General score from Qwen3.8 model card.

Category

Instruction Following

Max Score

—

Score Type

percent

Active

Yes

Model Scores

Scores for IFBench

Model Score Date Verified
Fable-5
67.89%
18.08.2026 Unverified
GPT-5.6 Sol
83.19%
18.08.2026 Unverified
Claude Opus 4.8
65.72%
18.08.2026 Unverified
Qwen3.7-Max
93.84%
18.08.2026 Unverified
Qwen3.8-2.4T-A95B
100.00%
18.08.2026 Verified
Qwen3.5-4B
60.73%
01.02.2026 Verified
Muse-Glimmer-30B
90.35%
01.08.2026 Verified
Gemma4-31B-it
88.69%
01.08.2026 Unverified
Qwen3.6-27B
80.03%
01.08.2026 Unverified
Qwen3.5-122B-A10B
88.85%
01.02.2026 Verified
Qwen3.5-35B-A3B
79.03%
01.02.2026 Unverified
Qwen3.5-27B
89.52%
01.02.2026 Unverified
Qwen3-235B-A22B
48.25%
01.02.2026 Unverified
GPT-5-mini 2025-08-07
87.69%
01.02.2026 Unverified
Qwen3.8-Flash-Next
97.50%
26.08.2026 Verified
Qwen3.8-27B
94.51%
26.08.2026 Unverified
Qwen3.7-Plus
93.84%
26.08.2026 Unverified
DeepSeek-V4-Flash-0731
94.01%
26.08.2026 Unverified
Claude-Opus-4.6 (Max)
66.22%
26.08.2026 Unverified
Spark-X2.5-4B
87.02%
— Verified
Spark-X2.5-1.7B
72.55%
— Verified
Qwen3.5-9B
69.55%
— Verified
Qwen3.5-2B
30.95%
— Verified
Gemma4-12B
84.53%
— Verified
Gemma4-E4B
35.44%
— Verified
Gemma4-E2B
22.70
— Verified
MiniCPM5-2B
72.55%
— Verified
granite-4.2-3B
83.69%
— Verified
Nemotron-3-Nano-4B
59.23%
— Verified
LFM2.5-8B-A1B
47.09%
— Verified
LFM2.5-2.6B
60.40%
— Verified