Share your thoughts, 1 month free Claude Pro on usSee more

Instruction Following on IFBench (Accuracy)

76.5Accuracy

Qwen3.5-27B

Updated 3d ago

Evaluation Results

Method	Links
Qwen3.5-27B 2026.04		76.5
GPT-5 mini 2026.04		74
Qwen3.5-122B-A10B 2026.04		73.77
Nemotron 3 Super 2026.04		72.56
GPT-OSS-120B 2026.04		68.32
gpt-oss-puzzle-88B 2026.02		67.77
K-EXAONE-236B-A23B 2026.04		67.3
gpt-oss-puzzle-88B 2026.02		67.01
gpt-oss-120B 2026.02		65.76
gpt-oss-puzzle-88B 2026.02		65.56
gpt-oss-120B 2026.02		64.46
gpt-oss-puzzle-88B 2026.02		63.35
EXAONE 4.5 33B 2026.04		62.6
Qwen3-VL-235B-A22B 2026.04		59.2
gpt-oss-120B 2026.02		58.67
gpt-oss-120B 2026.02		56.55
gpt-oss-puzzle-88B 2026.02		56.38
gpt-oss-puzzle-88B 2026.02		56.21
gpt-oss-120B 2026.02		44.13
gpt-oss-120B 2026.02		43.71
Qwen3-4B 2025.11		30.28
LFM2-8B-A1B 2025.11		25.85
Gemma-3-4B 2025.11		23.53
LFM2-2.6B 2025.11		22.19
Llama-3.2-3B 2025.11		20.78
OLMo3-7B-Think-DPO + GRPO (FLIP) 2026.02		20.5
OLMo3-7B-Think-DPO + GRPO (FLIP) 2026.02		20.1
OLMo3-7B-Think 2026.02		19.8
OLMo3-7B-Think-DPO + GRPO 2026.02		19.6
OLMo3-7B-Think-DPO + GRPO 2026.02		19.5
OLMo3-7B-Think-DPO 2026.02		18.8
Granite-4.0-H 2025.11		18.37
SmolLM3-3B 2025.11		17.93