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.
發布日期
2026-03-02
參數規模
4.0B
上下文長度
—
支援模態
—
能力雷達圖
19
general
14
coding
71
reasoning
40
science估算
70
agents
50
multimodal
Science 在缺少專門科學評測時使用推理能力代理估算。
排行榜排名
基準測試分數 (LLM Stats)
Agents
t2-bench
79.9%自報
BFCL-V4
50.3%自報
VITA-Bench
22.0%自報
DeepPlanning
17.6%自報
Biology
GPQA
76.2%自報
Chemistry
SuperGPQA
52.9%自報
Communication
Multi-Challenge
49.0%自報
Finance
MMLU-Pro
79.1%自報
MMLU-ProX
71.5%自報
General
IFEval
89.8%自報
MMLU-Redux
88.8%自報
C-Eval
85.1%自報
Global PIQA
78.9%自報
MAXIFE
78.0%自報
MMMLU
76.1%自報
Include
71.0%自報
IFBench
59.2%自報
LiveCodeBench v6
55.8%自報
NOVA-63
54.3%自報
LongBench v2
50.0%自報
Language
WMT24++
66.6%自報
Long Context
AA-LCR
57.0%自報
Math
HMMT25
76.8%自報
HMMT 2025
74.0%自報
PolyMATH
51.1%自報
AA 評測指數
Intelligence Index22.6
Coding Index13.7
Tau20.9
Gpqa0.7
Ifbench0.3
Lcr0.3
Scicode0.2
Terminalbench Hard0.1
Hle0.1
LLM Stats 分類評分
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
定價
輸入價格$0.03 / 1M tokens
輸出價格$0.15 / 1M tokens
混合價格(3:1)$0.06 / 1M tokens
速度
Tokens/秒216.4 tokens/s
首Token延遲0.25s
首回答延遲0.25s
可用提供商
(LS 內部計價單位)暫無提供商資料