Qwen3.5 9B (Non-reasoning)
AlibabaQwen开源权重Apache 2.0 · 商用许可
描述
Qwen3.5-9B is a 9 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 across knowledge, reasoning, coding, and multilingual tasks.
发布日期
2026-03-02
参数规模
9.0B
上下文长度
262K
支持模态
image, text, video
能力雷达图
18
general
24
coding
79
reasoning
47
science估算
70
agents
60
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
79.1%自报
BFCL-V4
66.1%自报
VITA-Bench
29.8%自报
DeepPlanning
18.0%自报
Biology
GPQA
81.7%自报
Chemistry
SuperGPQA
58.2%自报
Communication
Multi-Challenge
54.5%自报
Finance
MMLU-Pro
82.5%自报
MMLU-ProX
76.3%自报
General
IFEval
91.5%自报
MMLU-Redux
91.1%自报
C-Eval
88.2%自报
MAXIFE
83.4%自报
Global PIQA
83.2%自报
MMMLU
81.2%自报
Include
75.6%自报
LiveCodeBench v6
65.6%自报
IFBench
64.5%自报
NOVA-63
55.9%自报
LongBench v2
55.2%自报
Language
WMT24++
72.6%自报
Long Context
AA-LCR
63.0%自报
Math
HMMT 2025
83.2%自报
HMMT25
82.9%自报
PolyMATH
57.3%自报
AA 评测指数
Coding Index23.5
Intelligence Index20.3
Tau20.9
Gpqa0.8
Lcr0.4
Ifbench0.4
Scicode0.3
Terminalbench V2 10.2
Terminalbench Hard0.2
Hle0.1
Tau Banking0.0
LLM Stats 分类评分
Instruction Following80
Language80
Math80
General80
Biology80
Legal70
Physics70
Reasoning70
Structured Output70
Finance70
Healthcare70
Chemistry70
Tool Calling70
Long Context60
Multimodal60
Spatial Reasoning60
Economics60
Vision60
Agents50
Communication50
定价
输入价格免费
输出价格免费
混合价格(3:1)免费
速度
Tokens/秒0.0
首Token延迟0.00s
首回答延迟0.00s
供应商价格排行
供应商价格排行
4 个供应商
最便宜: NanoGPT最贵: Mixlayer
供应商输入输出
1NanoGPT最便宜
$0.05
$0.15
2Kilo Gateway
$0.05
$0.15
3OpenRouter
$0.1
$0.15
4Mixlayer
$0.1
$0.4
比较该模型在不同 API 供应商之间的定价。