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
基准测试分数 (LLM Stats)
Agents
t2-bench
79.9%自报
BFCL-V4
50.3%自报
VITA-Bench
22.0%自报
DeepPlanning
17.6%自报
Biology
GPQA
76.2%自报
Chemistry
SuperGPQA
52.9%自报
Communication
Multi-Challenge
49.0%自报
Finance
MMLU-Pro
79.1%自报
MMLU-ProX
71.5%自报
General
IFEval
89.8%自报
MMLU-Redux
88.8%自报
C-Eval
85.1%自报
Global PIQA
78.9%自报
MAXIFE
78.0%自报
MMMLU
76.1%自报
Include
71.0%自报
IFBench
59.2%自报
LiveCodeBench v6
55.8%自报
NOVA-63
54.3%自报
LongBench v2
50.0%自报
Language
WMT24++
66.6%自报
Long Context
AA-LCR
57.0%自报
Math
HMMT25
76.8%自报
HMMT 2025
74.0%自报
PolyMATH
51.1%自报
AA 评测指数
Intelligence Index22.6
Coding Index13.7
Tau20.9
Gpqa0.7
Ifbench0.3
Lcr0.3
Scicode0.2
Terminalbench Hard0.1
Hle0.1
LLM Stats 分类评分
Biology80
Language80
Structured Output70
Tool Calling70
Finance70
General70
Healthcare70
Instruction Following70
Legal70
Math70
Physics70
Chemistry60
Reasoning60
Spatial Reasoning50
Vision50
Communication50
Economics50
Long Context50
Multimodal50
Agents40
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