跳轉到主要內容

Qwen3 VL 235B A22B Instruct

AlibabaQwen開源權重Apache 2.0 · 商用許可

描述

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.

發布日期
2025-09-23
參數規模
236.0B
上下文長度
262K
支援模態
image, text, video

能力雷達圖

90
general
0
coding
80
reasoning
51
science估算
70
agents
90
multimodal

缺少專門科學評測時,Science 由 LLM Stats 科學得分或推理能力估算。

排行榜排名

領域#排名分數來源
多模態榜67
54.0
LS

基準測試分數 (LLM Stats)

(LLM Stats (zeroeval))

Chat

MM-MT-Bench8.50 / 100自報
IFEvalGoogle Research (2023)87.8%自報
Multi-IF76.3%自報

Creativity

Creative Writing v386.5%自報

Factuality

SimpleQA51.9%自報

General

MMLU88.8%自報
MultiPL-E86.1%自報
CSimpleQA83.4%自報
Include80.0%自報
Arena-Hard v277.4%自報
BFCL-v367.7%自報
AndroidWorld_SR63.7%自報

Language

MMLU-Redux92.2%自報
MMLU-Pro81.8%自報
MMLU-ProX77.8%自報

Math

MathVista-Mini84.9%自報
LiveBench 2024112574.8%自報
AIME 202574.7%自報
MathVision66.5%自報
HMMT2557.4%自報

Multimodal

VideoMME w/o sub.79.2%自報
VideoMMMU74.7%自報
OSWorld66.7%自報

Reasoning

CharXiv-R62.1%自報
LiveCodeBench v561.4%自報
SuperGPQA60.4%自報
LiveCodeBench v654.3%自報

Video

MLVU84.3%自報
CharadesSTA64.8%自報

Vision

DocVQAtest97.1%自報
ScreenSpot95.4%自報
OCRBench92.0%自報
MMBench-V1.189.9%自報
AI2D89.7%自報
InfoVQAtest89.2%自報
CC-OCR82.2%自報
RealWorldQA79.3%自報
MMMUval78.7%自報
MMStar78.4%自報
MuirBench72.8%自報
BLINK70.7%自報
MMMU-Pro68.1%自報
LVBench67.7%自報
OCRBench-V2 (en)67.1%自報
Hallusion Bench63.2%自報
ScreenSpot Pro62.0%自報
OCRBench-V2 (zh)61.8%自報
ERQA51.3%自報
ODinW48.6%自報

Writing

WritingBench85.5%自報

AA 評測指數

(Artificial Analysis)

暫無 AA 評測資料

LLM Stats 分類評分

(LLM Stats (zeroeval))
Chat
3
Communication
3
Multimodal
1
General
90
Instruction Following
80
Language
80
Legal
80
Long Context
80
Math
80
Structured Output
80
Finance
80
Grounding
80
Healthcare
80
Creativity
80
Text-to-image
80
Video
80
Writing
80
Image To Text
70
Reasoning
70
Spatial Reasoning
70
3d
70
Agents
70
Tool Calling
70
Vision
70
Physics
60
Chemistry
60
Economics
60
Factuality
50

定價

輸入價格$0.215 / 1M tokens
輸出價格$1.39 / 1M tokens
混合價格(3:1)$0.37 / 1M tokens
快取讀取價格$0.1 / 1M tokens

速度

Tokens/秒43.3
首Token延遲1.22s
首回答延遲1.22s

供應商價格排行

供應商價格排行

10 個供應商

最便宜: DeepInfra最貴: OrcaRouter
供應商輸入輸出
1DeepInfra最便宜
$0.2
$0.88
2DevPass (LLM Gateway)
$0.2
$0.88
3OpenRouter
$0.21
$1.9
4Kilo Gateway
$0.21
$1.9
5Merge Gateway
$0.21
$1.9
6Alibaba主要
$0.215
$1.39
7NanoGPT
$0.3
$1.9
8Helicone
$0.3
$1.5
9Prime Inference
$0.3
$1.5
10OrcaRouter
$0.4
$1.6

比較該模型在不同 API 供應商之間的定價。

外部連結