Saltar al contenido principal

Qwen3 VL 30B A3B Instruct

AlibabaQwenOpen WeightApache 2.0 · Uso Comercial

Descripción

Qwen3-VL is a large multimodal model that unifies vision, language, and reasoning to achieve human-level perception and cognition across text, images, and video. Built on a 235B-parameter architecture, it integrates early joint training of visual and textual modalities for strong language grounding. The model supports up to a 1 million-token context window and excels at visual understanding, spatial reasoning, long video comprehension, and tool-based interaction. It can generate code from images, perform precise 2D/3D object grounding, and operate digital interfaces like a visual agent. The “Instruct” version rivals Gemini 2.5 Pro in perception benchmarks, while the “Thinking” version leads in multimodal reasoning and STEM tasks. With multilingual OCR, creative writing, and fine-grained scene interpretation, Qwen3-VL establishes a new open-source frontier for integrated vision-language intelligence.

Fecha de lanzamiento
2025-10-03
Parámetros
31.0B
Longitud del contexto
262K
Modalidades
image, text, video

Radar de capacidades

80
general
0
coding
70
reasoning
60
scienceest.
70
agents
100
multimodal

Science se estima a partir de las puntuaciones científicas de LLM Stats o del razonamiento cuando no hay benchmarks científicos dedicados.

Rankings

Dominio#PosiciónPuntuaciónFuente
Ranking multimodal110
38.0
LS

Puntuaciones de benchmarks (LLM Stats)

(LLM Stats (zeroeval))

Chat

MM-MT-Bench8.10 / 100Aut.
IFEvalGoogle Research (2023)85.8%Aut.
Multi-IF66.1%Aut.

Creativity

Creative Writing v384.6%Aut.

Factuality

SimpleQA27.0%Aut.

General

MMLU85.0%Aut.
MLVU-M81.3%Aut.
Include71.6%Aut.
BFCL-v366.3%Aut.
Arena-Hard v258.5%Aut.

Language

MMLU-Redux88.4%Aut.
MMLU-Pro77.8%Aut.
MMLU-ProX70.9%Aut.

Math

MathVista-Mini80.1%Aut.
AIME 202569.3%Aut.
LiveBench 2024112565.4%Aut.
MathVision60.2%Aut.
HMMT2550.6%Aut.
PolyMATH44.3%Aut.

Multimodal

Video-MME74.5%Aut.
VideoMMMU68.7%Aut.
OSWorld30.3%Aut.

Reasoning

CharXiv-D85.5%Aut.
GPQANYU + Cohere + Anthropic (2023)70.4%Aut.
SuperGPQA53.1%Aut.
CharXiv-R48.9%Aut.
LiveCodeBench v642.6%Aut.

Video

CharadesSTA63.5%Aut.

Vision

DocVQAtest95.0%Aut.
ScreenSpot94.7%Aut.
OCRBench90.3%Aut.
MMBench-V1.187.0%Aut.
AI2D85.0%Aut.
InfoVQAtest82.0%Aut.
CC-OCR80.7%Aut.
MMMU (val)74.2%Aut.
RealWorldQA73.7%Aut.
MVBench72.3%Aut.
MMStar72.1%Aut.
BLINK67.7%Aut.
OCRBench-V2 (en)63.2%Aut.
MuirBench62.9%Aut.
LVBench62.5%Aut.
Hallusion Bench61.5%Aut.
ScreenSpot Pro60.5%Aut.
MMMU-Pro60.4%Aut.
OCRBench-V2 (zh)57.8%Aut.
ODinW47.5%Aut.
ERQA43.0%Aut.

Writing

WritingBench82.6%Aut.

Índices de evaluación AA

(Artificial Analysis)

No hay datos de evaluación AA disponibles

Puntuaciones por categoría LLM Stats

(LLM Stats (zeroeval))
Chat
3
Communication
3
Multimodal
100
Instruction Following
80
Language
80
Structured Output
80
General
80
Grounding
80
Creativity
80
Text-to-image
80
Writing
80
Image To Text
70
Legal
70
Math
70
Reasoning
70
Spatial Reasoning
70
Finance
70
Healthcare
70
3d
70
Biology
70
Tool Calling
70
Video
70
Vision
70
Long Context
60
Physics
60
Chemistry
60
Agents
50
Economics
50
Factuality
30

Precios

Precio de entrada$0.35 / 1M tokens
Precio de salida$0.6 / 1M tokens
Precio mixto (3:1)$0.2625 / 1M tokens

Velocidad

Tokens/seg43.9
Retraso del primer token2.38s
Tiempo hasta la respuesta2.38s

Ranking de Precios por Proveedor

Ranking de Precios por Proveedor

9 proveedores

Más barato: DeepInfraMás caro: Alibaba
ProveedorEntradaSalida
1DeepInfraMás barato
$0.15
$0.6
2OpenRouter
$0.15
$0.6
3Kilo Gateway
$0.15
$0.6
4DevPass (LLM Gateway)
$0.15
$0.6
5Merge Gateway
$0.15
$0.6
6Novita AI
$0.2
$0.7
7Prime Inference
$0.2
$0.7
8Phala
$0.2
$0.7
9AlibabaPRINCIPAL
$0.35
$0.6

Comparar precios entre diferentes proveedores de API para este modelo.

Fuentes externas