8.7B- Context
- 66Ktokens
- Released
- 11mo ago
- Size
- 9Bparameters
AutoGLM-Phone-9B-Multilingual is an open-weights, MIT-licensed model from Z.ai, with GLM-4.1V-9B-Base as its base model and an architecture identical to GLM-4.1V-9B-Thinking. It is built for Phone Agent, a mobile assistant framework that reads what is on a phone screen, plans the steps needed, and carries out tasks a user describes in natural language, asking for confirmation before sensitive actions. Z.ai limits it to research and educational use.
- Multimodal model
- Open weights · 9B
- Accepts images
How it compares
Additional source evidence
- Source model
- zhipuai/autoglm-phone-9b-multilingual
- Source family
- glm
- Last updated
- 2025-12-10
- Model context
- 66K
- Model output limit
- 66K
- Open weights
- Yes
- File attachments
- Yes
- Reasoning
- No
- Tool calling
- No
- Temperature control
- Yes
Downloadable weights
Complete source record (JSON)
{
"id": "zhipuai/autoglm-phone-9b-multilingual",
"name": "AutoGLM Phone 9B Multilingual",
"description": "Multilingual visual model for operating smartphone interfaces and mobile agents",
"family": "glm",
"last_updated": "2025-12-10",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": true,
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"limit": {
"context": 65536,
"output": 65536
},
"release_date": "2025-12-10",
"open_weights": true,
"license": "MIT",
"weights": [
{
"label": "Hugging Face",
"url": "https://huggingface.co/zai-org/AutoGLM-Phone-9B-Multilingual"
}
]
}Frequently asked questions
Running locally
At Q4 AutoGLM Phone 9B Multilingual needs an estimated 6.8 GB of VRAM, at Q8 about 11.5 GB, and at FP16 about 21.2 GB. These figures are estimates from the 9B parameter count, covering weights plus about 15% and 0.5 GB for a short context.
Yes. A 24 GB GPU fits FP16 (estimated 21.2 GB), a 16 GB or 12 GB GPU fits Q8 (estimated 11.5 GB), and an 8 GB GPU fits Q4 (estimated 6.8 GB).
Yes. At Q4 it needs an estimated 6.8 GB, and assuming macOS lets the GPU use about 75% of unified memory, the smallest common Mac memory size that fits Q4 is 16 GB.
Privacy and data
Yes. AutoGLM Phone 9B Multilingual has open weights, so you can self-host it for full control of your data.
Capabilities and specs
66K tokens (65,536) for the context window, and maximum output is also 66K tokens (65,536).
Yes, it can read images. Its input modalities are text and image, and audio, video or PDF inputs are not listed.
No. AutoGLM Phone 9B Multilingual does not support tool calling (function calling).
No. AutoGLM Phone 9B Multilingual does not support a reasoning or thinking mode.
Quality and benchmarks
Molmo 2 8B by Ai2, Olmo 3.1 32B Instruct by Ai2 and ERNIE 4.5 VL 28B A3B Thinking by Baidu. They were matched on model kind and release date, with no comparable price.
Molmo 2 8BAutoGLM Phone 9B Multilingual vs Molmo 2 8BOlmo 3.1 32B InstructAutoGLM Phone 9B Multilingual vs Olmo 3.1 32B InstructERNIE 4.5 VL 28B A3B ThinkingAutoGLM Phone 9B Multilingual vs ERNIE 4.5 VL 28B A3B Thinking
Licensing and access
Open weights. AutoGLM Phone 9B Multilingual is released with open weights under the MIT license.
Yes. The MIT license is permissive and allows commercial use.
The weights are available on Hugging Face in the repository zai-org/AutoGLM-Phone-9B-Multilingual.
Basics
Z.ai released AutoGLM Phone 9B Multilingual on December 10, 2025.
Yes, it has open weights.
Generated from Vertical Tabs data, updated November 4, 2026.
Discussion
No comments yet
Be the first to share your thoughts.