Saltar al contenido principal
AITier

Qwen3.5 4B (Non-reasoning)

AlibabaQwenOpen WeightApache 2.0 · Commercial OK

Descripción

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.

Fecha de lanzamiento
2026-03-02
Parámetros
4.0B
Longitud del contexto
—
Modalidades
—

Radar de capacidades

19
general
14
coding
71
reasoning
40
scienceest.
70
agents
50
multimodal

Science usa un proxy de razonamiento cuando los benchmarks científicos dedicados no están disponibles.

Rankings

Dominio#PosiciónPuntuaciónFuente
Agents & Tools78
47.0
LS
Code Ranking292
26.0
AA
General Ranking222
46.0
AA
Science281
39.0
AA

Puntuaciones de benchmarks (LLM Stats)

Agents

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

Biology

GPQA76.2%Aut.

Chemistry

SuperGPQA52.9%Aut.

Communication

Multi-Challenge49.0%Aut.

Finance

MMLU-Pro79.1%Aut.
MMLU-ProX71.5%Aut.

General

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

Language

WMT24++66.6%Aut.

Long Context

AA-LCR57.0%Aut.

Math

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

Índices de evaluación AA

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

Puntuaciones por categoría LLM Stats

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

Precios

Precio de entrada$0.03 / 1M tokens
Precio de salida$0.15 / 1M tokens
Precio mixto (3:1)$0.06 / 1M tokens

Velocidad

Tokens/seg216.4 tokens/s
Retraso del primer token0.25s
Tiempo hasta la respuesta0.25s

Proveedores disponibles

(Unidades internas LS)

No hay datos de proveedores disponibles

Fuentes externas

LLM Stats