Ministral 14B
Trinity Nano Preview
- Context
- 262Ktokens
- Released
- 8mo ago
- Size
- 9Bparameters
Pre-trained Qwen3.5 base checkpoint for fine-tuning and research, without instruction tuning
- Multimodal model
- Open weights · 9B
- Accepts images, video
How it compares
Size class
Small9BAn open-weight model of 5 to 25 billion parameters, the size that runs on a single consumer GPU.
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/qwen3.5-9b-base
- Source family
- qwen
- Last updated
- 2026-02-26
- Model context
- 262K
- 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": "alibaba/qwen3.5-9b-base",
"name": "Qwen3.5 9B Base",
"description": "Pre-trained Qwen3.5 base checkpoint for fine-tuning and research, without instruction tuning",
"family": "qwen",
"last_updated": "2026-02-26",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": true,
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"limit": {
"context": 262144,
"output": 65536
},
"release_date": "2026-02-26",
"open_weights": true,
"license": "Apache-2.0",
"weights": [
{
"label": "Hugging Face",
"url": "https://huggingface.co/Qwen/Qwen3.5-9B-Base"
}
]
}Frequently asked questions
Running locally
Estimated VRAM is 6.8 GB at Q4, 11.5 GB at Q8 and 21.2 GB at FP16, based on the 9B parameter count. These estimates cover weights plus about 15% and 0.5 GB for a short context.
Yes. It fits at FP16 on a 24 GB GPU (estimated 21.2 GB), at Q8 on 16 GB and 12 GB GPUs (estimated 11.5 GB) and at Q4 on an 8 GB GPU (estimated 6.8 GB).
Yes, at Q4 it needs an estimated 6.8 GB. Assuming macOS lets the GPU use about 75% of unified memory, the smallest common Mac memory size that fits is 16 GB.
Privacy and data
Yes, Qwen3.5 9B Base has open weights, so it can be self-hosted for full control of your data.
Capabilities and specs
The context window is 262K tokens (262,144) and the maximum output is 66K tokens (65,536).
It can read images and video, but not audio or PDFs. Its input modalities are text, image and video.
No, tool calling (function calling) is not supported.
No, a reasoning or thinking mode is not supported.
Quality and benchmarks
The closest alternatives, matched on model kind and release date with no comparable price, are Olmo Hybrid 7B Instruct by Ai2, Olmo Hybrid 7B Think by Ai2 and Nemotron Cascade 2 30B A3B by NVIDIA.
Olmo Hybrid 7B InstructQwen3.5 9B Base vs Olmo Hybrid 7B InstructOlmo Hybrid 7B ThinkQwen3.5 9B Base vs Olmo Hybrid 7B ThinkNemotron Cascade 2 30B A3BQwen3.5 9B Base vs Nemotron Cascade 2 30B A3B
Versions and lineage
They have the same context window of 262K tokens (262,144). Qwen3.5 9B Base was released on February 26, 2026, while Qwen3.5 0.8B was released on March 2, 2026.
Licensing and access
Open weights, released under the Apache-2.0 license.
Yes, the Apache 2.0 license is permissive and allows commercial use.
Basics
Qwen made Qwen3.5 9B Base, and it was released on February 26, 2026.
Yes, Qwen3.5 9B Base is open weights.
Generated from Vertical Tabs data, updated November 10, 2026.
Discussion
No comments yet
Be the first to share your thoughts.