Evaluation overview

Overall rank

#108

64.7 points · 180 models

Open-weight rank

#48

70.3 points · 109 models

Data coverage

7 items

1 source · updated through 2026-02-27

Domain performance

Research#104 / 174

74.9

Evaluation highlights

All evaluations

MMLU-Redux

Reasoning · Qwen3.5-9B Technical Report

88.8%

43/ 102

IFEval strict prompt

Instruction following · Qwen3.5-9B Technical Report

89.8%

14/ 55

C-Eval

Chinese · Qwen3.5-9B Technical Report

85.1%

29/ 64

MMLU-Pro

Reasoning · Qwen3.5-9B Technical Report

79.1%

47/ 95

SuperGPQA

Reasoning · Qwen3.5-9B Technical Report

52.9%

33/ 65

GPQA-Diamond

Reasoning · Qwen3.5-9B Technical Report

76.2%

69/ 126

Basic information

Context (input / output): 262K / 33K
Parameters: 4B
Release date: 2026-02
License: Apache 2.0
Input modalities: Text · Image
Output modalities: Text

Availability

vLLM · BF16Deployment options

BF16 · 10 GB VRAM

Official site HuggingFace

Qwen3.5-4BOpen

Evaluation overview

Domain performance

Evaluation highlights

Basic information

Availability

Qwen3.5-4BOpen

Evaluation overview

Domain performance

Evaluation highlights

Basic information

Availability

Evaluation overview

Domain performance

Evaluation highlights

Basic information

Availability

Related links

Evaluation overview

Domain performance

Evaluation highlights

Basic information

Availability

Related links