- Context
- 131Ktokens
- Released
- 1y ago
- Size
- 235Bparameters
Qwen3 VL 235B A22B Thinking is Alibaba's open-weights Thinking variant of Qwen3-VL-235B-A22B, the flagship model of the new Qwen3-VL vision-language series. Alibaba's launch announcement centers the model on stronger visual perception, deeper reasoning, and agentic tool use. The Thinking version is built for multimodal reasoning: Alibaba highlights STEM and math problem solving, agent use of computer and mobile interfaces, visual coding, document and video understanding, and text recognition.
- Multimodal model
- Open weights · 235B
- Accepts images
- Reasoning
- Tools
How it compares
Size class
Large235BAn open-weight model of 100 billion parameters or more, which needs a multi-GPU server.
Open models of a similar size from other labs
Vs. your stack
Add the models you use to see how this one compares.
Sign inAdditional source evidence
- Source model
- alibaba/qwen3-vl-235b-a22b-thinking
- Source family
- qwen
- Last updated
- 2025-09-23
- Knowledge cutoff
- 2025-03-31
- Model context
- 131K
- Model output limit
- 33K
- Open weights
- Yes
- File attachments
- Yes
- Reasoning
- Yes
- Tool calling
- Yes
- Temperature control
- Yes
Downloadable weights
Complete source record (JSON)
{
"id": "alibaba/qwen3-vl-235b-a22b-thinking",
"name": "Qwen3 VL 235B A22B Thinking",
"description": "Qwen vision-language thinking model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"last_updated": "2025-09-23",
"knowledge": "2025-03-31",
"attachment": true,
"reasoning": true,
"tool_call": true,
"temperature": true,
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"limit": {
"context": 131072,
"output": 32768
},
"release_date": "2025-09-23",
"open_weights": true,
"weights": [
{
"label": "Hugging Face",
"url": "https://huggingface.co/Qwen/Qwen3-VL-235B-A22B-Thinking"
}
]
}Frequently asked questions
Running locally
Qwen3 VL 235B A22B Thinking needs an estimated 164.3 GB of VRAM at Q4, 287.6 GB at Q8 and 541 GB at FP16. These are estimates from the 235B parameter count, covering weights plus about 15% and 0.5 GB for a short context.
No. A 24 GB, 16 GB, 12 GB or 8 GB GPU does not fit Qwen3 VL 235B A22B Thinking at Q4, Q8 or FP16, according to estimates.
Yes, at Q4 it needs an estimated 164.3 GB. Assuming macOS lets the GPU use about 75% of unified memory, the smallest common Mac memory size that fits Q4 is 256 GB (estimate).
Privacy and data
Yes. Qwen3 VL 235B A22B Thinking has open weights, so you can self-host it for full control of your data.
Capabilities and specs
The context window is 131K tokens (131,072) and the maximum output is 33K tokens (32,768).
Its input modalities are text and image. Audio, video and PDF input are not listed.
Yes, tool calling (function calling) is supported.
Yes, a reasoning mode is supported.
The knowledge cutoff is 2025-03-31.
Quality and benchmarks
On Arena Vision it ranks 102 (Arena's own rank) with a score of 1188. On Arena Text it ranks 173 (Arena's own rank) with a score of 1395.
Matched on model kind and release date, with no comparable price: Gemma-SEA-LION-v4-27B-IT by AI Singapore, Magistral Small 1.2 by Mistral AI and Granite-4.0-H-Micro by IBM.
Gemma-SEA-LION-v4-27B-ITQwen3 VL 235B A22B Thinking vs Gemma-SEA-LION-v4-27B-ITMagistral Small 1.2Qwen3 VL 235B A22B Thinking vs Magistral Small 1.2Granite-4.0-H-MicroQwen3 VL 235B A22B Thinking vs Granite-4.0-H-Micro
Versions and lineage
Both models have a context window of 131K tokens (131,072). Qwen3 VL 235B A22B Thinking was released on September 23, 2025, while Qwen3 235B-A22B was released on April 28, 2025.
Qwen3 235B-A22BQwen3 VL 235B A22B Thinking vs Qwen3 235B-A22B
Licensing and access
Qwen3 VL 235B A22B Thinking has open weights.
Basics
Qwen made Qwen3 VL 235B A22B Thinking, and it was released on September 23, 2025.
Qwen3 VL 235B A22B Thinking has open weights.
Generated from Vertical Tabs data, updated November 11, 2026.
More in Qwen
Similar models
Gemma-SEA-LION-v4-27B-ITGemma-SEA-LION-v4-27B-IT
24BMagistral Small 1.2
Granite-4.0-H-Micro
Granite-4.0-H-Small
21.8BERNIE 4.5 21B A3B Thinking
PaddleOCR VL
959MPaddleOCR VL
959MPaddleOCR-VL
8.3BolmOCR 2
685BDeepSeek-V3.1
GPT OSS Safeguard 120BGPT OSS Safeguard 120B
9BNemotron Nano 9B v2
8B
Discussion
No comments yet
Be the first to share your thoughts.