跳转到主要内容

Qwen3.5 4B (Non-reasoning)

AlibabaQwenOpen WeightApache 2.0 · Commercial OK

描述

Qwen3.5-4B is a 4 billion parameter vision-language model using Gated DeltaNet hybrid architecture with a 3:1 ratio of linear attention to full softmax attention. It supports 262K native context length and delivers strong performance for its size across knowledge, reasoning, coding, and multilingual tasks.

发布日期
2026-03-02
参数规模
4.0B
上下文长度
支持模态

能力雷达图

19
general
14
coding
71
reasoning
40
science估算
70
agents
50
multimodal

Science 在缺少专门科学评测时使用推理能力代理估算。

排行榜排名

领域#排名分数来源
智能体与工具78
47.0
LS
代码能力榜292
26.0
AA
通用能力榜222
46.0
AA
科学能力281
39.0
AA

基准测试分数 (LLM Stats)

Agents

t2-bench79.9%自报
BFCL-V450.3%自报
VITA-Bench22.0%自报
DeepPlanning17.6%自报

Biology

GPQA76.2%自报

Chemistry

SuperGPQA52.9%自报

Communication

Multi-Challenge49.0%自报

Finance

MMLU-Pro79.1%自报
MMLU-ProX71.5%自报

General

IFEval89.8%自报
MMLU-Redux88.8%自报
C-Eval85.1%自报
Global PIQA78.9%自报
MAXIFE78.0%自报
MMMLU76.1%自报
Include71.0%自报
IFBench59.2%自报
LiveCodeBench v655.8%自报
NOVA-6354.3%自报
LongBench v250.0%自报

Language

WMT24++66.6%自报

Long Context

AA-LCR57.0%自报

Math

HMMT2576.8%自报
HMMT 202574.0%自报
PolyMATH51.1%自报

AA 评测指数

Intelligence Index
22.6
Coding Index
13.7
Tau2
0.9
Gpqa
0.7
Ifbench
0.3
Lcr
0.3
Scicode
0.2
Terminalbench Hard
0.1
Hle
0.1

LLM Stats 分类评分

Biology
80
Language
80
Structured Output
70
Tool Calling
70
Finance
70
General
70
Healthcare
70
Instruction Following
70
Legal
70
Math
70
Physics
70
Chemistry
60
Reasoning
60
Spatial Reasoning
50
Vision
50
Communication
50
Economics
50
Long Context
50
Multimodal
50
Agents
40

定价

输入价格$0.03 / 1M tokens
输出价格$0.15 / 1M tokens
混合价格(3:1)$0.06 / 1M tokens

速度

Tokens/秒216.4 tokens/s
首Token延迟0.25s
首回答延迟0.25s

可用提供商

(LS 内部计价单位)

暂无提供商数据

外部链接