UI-TARS-1.5-7B needs about 5.7 GB of memory at Q4_K_M (a 4.8 GB file) with an 8K context window, so it fits on a 6 GB GPU. This repository has 27 quants, from 2.7 GB (IQ2_XXS) to 15.2 GB (BF16). Longer context needs more: at 32K tokens the KV cache alone adds about 1.9 GB.
UI‑TARS‑1.5‑7B: 7.6B‑parameter multimodal model with vision support
UI‑TARS‑1.5‑7B is a 7.6 billion‑parameter model built on the qwen2vl architecture. It supports image‑to‑text tasks and is released under the Apache‑2.0 license. The model is available in multiple GGUF quantizations for efficient deployment.
Good at
- Vision‑to‑text generation
- Multimodal image understanding
- Low‑latency inference on GPUs
Limits
- Requires separate mmproj file for vision
- Large context window of 128k tokens may need more VRAM
- No text‑only inference without mmproj
Best for
- Image captioning on edge devices
- Rapid prototyping of multimodal chatbots
- Batch image analysis with llama.cpp
Who it’s for. Developers building multimodal applications who need a lightweight, open‑source vision‑enabled model. The model is suitable for research and production use where an Apache‑2.0 license is acceptable.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| IQ2_XXS | 2.7 GB | 3.6 GB | Fits | Get file → | |
| IQ2_XS | 2.8 GB | 3.8 GB | Fits | Get file → | |
| IQ2_S | 2.9 GB | 3.9 GB | Fits | Get file → | |
| Q2_K_S | 2.9 GB | 3.9 GB | Fits | Get file → | |
| IQ2_M | 3.0 GB | 4.0 GB | Fits | Get file → | |
| Q2_K_M | 3.3 GB | 4.2 GB | Fits | Get file → | |
| IQ3_XXS | 3.3 GB | 4.2 GB | Fits | Get file → | |
| IQ3_XS | 3.4 GB | 4.4 GB | Fits | Get file → | |
| IQ3_S | 3.6 GB | 4.5 GB | Fits | Get file → | |
| IQ3_M | 3.6 GB | 4.6 GB | Fits | Get file → | |
| Q3_K_S | 3.6 GB | 4.6 GB | Fits | Get file → | |
| Q3_K_M | 4.0 GB | 5.0 GB | Fits | Get file → | |
| IQ4_XS | 4.2 GB | 5.2 GB | Fits | Get file → | |
| Q4_0 | 4.3 GB | 5.3 GB | Fits | Get file → | |
| IQ4_NL | 4.4 GB | 5.4 GB | Fits | Get file → | |
| Q4_K_S | 4.6 GB | 5.6 GB | Fits | Get file → | |
| Q4_1 | 4.8 GB | 5.7 GB | Fits | Get file → | |
| Q4_K_M | 4.8 GB | 5.7 GB | Fits | Get file → | |
| Q5_0 | 5.2 GB | 6.2 GB | Fits | Get file → | |
| Q5_K_S | 5.5 GB | 6.4 GB | Fits | Get file → | |
| Q5_K_M | 5.5 GB | 6.5 GB | Fits | Get file → | |
| Q5_1 | 5.7 GB | 6.7 GB | Fits | Get file → | |
| Q6_K_M | 6.3 GB | 7.2 GB | Tight | Get file → | |
| Q8_0 | 8.1 GB | 9.1 GB | Too big | Get file → | |
| BF16 | 9.1 GB | 10.1 GB | Too big | Get file → | |
| F16 | 11.3 GB | 12.3 GB | Too big | Get file → | |
| BF16 | 15.2 GB | 16.2 GB | Too big | Get file → |
For 8 GB VRAM, use Q4_K_M (4.78 GB file, ~5.7 GB VRAM at 8k context). For 12 GB VRAM, Q5_K_M (5.53 GB file, ~6.5 GB VRAM). For 16 GB VRAM, Q6_K_M (6.25 GB file, ~7.2 GB VRAM).
What GPU do you need for UI-TARS-1.5-7B?
These figures add the file size, an fp16 KV cache at an 8K context window (worked out from this model’s own header) and about 0.5 GB for the runtime. They are estimates, so leave a little headroom, and expect to use less context or a smaller quant if you are right at the limit.
- 4 GB VRAM: the best quant that fits is IQ2_M (3.0 GB file, about 4.0 GB in use).
- 6 GB VRAM: the best quant that fits is Q4_K_M (4.8 GB file, about 5.7 GB in use).
- 8 GB VRAM: the best quant that fits is Q6_K_M (6.3 GB file, about 7.2 GB in use).
- 12 GB VRAM: the best quant that fits is BF16 (9.1 GB file, about 10.1 GB in use).
- 16 GB VRAM: the best quant that fits is F16 (11.3 GB file, about 12.3 GB in use).
- 24 GB VRAM: the best quant that fits is BF16 (15.2 GB file, about 16.2 GB in use).
Which UI-TARS-1.5-7B quant should you choose?
Q4_K_M is the usual starting point for UI-TARS-1.5-7B: it balances file size and quality well. Here it is a 4.8 GB file.
The Q8_0 file is 8.1 GB, about 1.7 times the size of Q4_K_M. It keeps the weights closer to the original, which only matters if you have the memory to spare.
If memory is tight, the smallest file is IQ2_XXS at 2.7 GB, with the quality trade-offs that come with fewer bits. Use it only when memory is very tight. Expect a clear drop in quality, which is usually worse on small models and easier to live with on large ones.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about UI-TARS-1.5-7B
How much VRAM does UI-TARS-1.5-7B need?
About 5.7 GB at Q4_K_M with an 8K context window. That is the 4.8 GB file, plus the KV cache and roughly 0.5 GB of runtime overhead. Smaller quants need less and longer context needs more.
Can I run UI-TARS-1.5-7B on an 8 GB GPU?
Yes, with Q6_K_M: the 6.3 GB file needs about 7.2 GB at 8K context. Larger quants in this repository do not fit in 8 GB.
What is the file size of UI-TARS-1.5-7B GGUF?
It depends on the quant: IQ2_XXS is 2.7 GB, IQ2_XS is 2.8 GB, IQ2_S is 2.9 GB, Q2_K_S is 2.9 GB, IQ2_M is 3.0 GB, Q2_K_M is 3.3 GB and 21 more.
What context length does UI-TARS-1.5-7B support?
The GGUF header lists a trained context length of 128,000 tokens (125K). The KV cache grows with the context you use: about 0.5 GB at 8K and 1.9 GB at 32K.
Where can I download UI-TARS-1.5-7B GGUF?
The files are on Hugging Face in the Mungert/UI-TARS-1.5-7B-GGUF repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.
What license does UI-TARS-1.5-7B use?
Hugging Face lists the license as apache-2.0. Licenses can differ between the quantized files and the original model, so check the repository and the base model before commercial use.
Can I run UI‑TARS‑1.5‑7B on 8 GB VRAM?
Yes, using Q4_K_M or Q4_K_S quantizations will fit within 8 GB VRAM with approximate 5.7 GB usage.
Do I need a separate mmproj file?
Yes, the vision component requires the mmproj file; the gguf file alone is insufficient for image tasks.
What license applies to this model?
The model is released under the Apache‑2.0 license, allowing commercial and non‑commercial use.