unsloth
Qwen2.5 VL 7B Instruct
Qwen2.5 VL 7B Instruct is a 7.62B dense qwen2vl model from unsloth. Configured context length is 128,000 tokens. License: apache-2.0. Quant sizes are the GGUF file bytes published on Hugging Face when a file exists.
Parameters7.6B
Context125K
Attentiongqa
Licenseapache-2.0
28 layers · 4 KV heads · head dim 128 · GGUF repo · config
Quant file sizes
| Quant | File size | Source |
|---|---|---|
| BF16 | 14.2 GB | Hugging Face file bytes |
| UD-Q8_K_XL | 9.49 GB | Hugging Face file bytes |
| Q8_0 | 7.54 GB | Hugging Face file bytes |
| UD-Q6_K_XL | 6.48 GB | Hugging Face file bytes |
| Q6_K | 5.82 GB | Hugging Face file bytes |
| UD-Q5_K_XL | 5.09 GB | Hugging Face file bytes |
| Q5_K_M | 5.07 GB | Hugging Face file bytes |
| Q5_K_S | 4.95 GB | Hugging Face file bytes |
| UD-Q4_K_XL | 4.46 GB | Hugging Face file bytes |
| Q4_K_M | 4.36 GB | Hugging Face file bytes |
| Q4_K_S | 4.15 GB | Hugging Face file bytes |
| IQ4_NL | 4.13 GB | Hugging Face file bytes |
| IQ4_XS | 3.94 GB | Hugging Face file bytes |
| Q4_1 | 4.54 GB | Hugging Face file bytes |
| Q4_0 | 4.14 GB | Hugging Face file bytes |
| UD-Q3_K_XL | 3.73 GB | Hugging Face file bytes |
| Q3_K_M | 3.55 GB | Hugging Face file bytes |
| Q3_K_S | 3.25 GB | Hugging Face file bytes |
| UD-IQ3_XXS | 2.95 GB | Hugging Face file bytes |
| UD-Q2_K_XL | 3.02 GB | Hugging Face file bytes |
| Q2_K | 5.74 GB | Hugging Face file bytes |
| UD-IQ2_XXS | 2.23 GB | Hugging Face file bytes |
| UD-IQ1_M | 2.05 GB | Hugging Face file bytes |
| UD-IQ1_S | 1.93 GB | Hugging Face file bytes |
On common machines
| Device | Verdict | Quant | Speed |
|---|---|---|---|
| GeForce RTX 4060 Ti 8GB | Tight fit | Q6_K | 24–32 |
| GeForce RTX 3060 12GB | Tight fit | UD-Q8_K_XL | 19–25 |
| GeForce RTX 4060 Ti 16GB | Runs | UD-Q8_K_XL | 15–20 |
| GeForce RTX 4090 | Runs | BF16 | 35–48 |
| GeForce RTX 5090 | Runs great | BF16 | 63–86 |
| Apple M4 Pro | Runs | BF16 | 9.6–13 |
| Apple M5 | Tight fit | UD-Q8_K_XL | 7.9–11 |
| CPU only · DDR5-5600 | Runs on CPU | BF16 | 3.1–4.3 |
Smallest devices that fit it fully
- GeForce RTX 3050 6GB — Tight fit, UD-Q4_K_XL, 18–25 tok/s
- GeForce RTX 2060 6GB — Tight fit, UD-Q4_K_XL, 37–50 tok/s
- GeForce GTX 1660 Super — Tight fit, UD-Q4_K_XL, 37–50 tok/s
- GeForce GTX 1660 Ti — Tight fit, UD-Q4_K_XL, 32–43 tok/s
- Intel Arc A380 — Tight fit, UD-Q4_K_XL, 20–28 tok/s
- GeForce RTX 5060 Ti 8GB — Tight fit, Q6_K, 37–50 tok/s
Run it
Commands for the GeForce RTX 4060 Ti 16GB estimate (UD-Q8_K_XL).
llama.cpp
llama-server -hf unsloth/Qwen2.5-VL-7B-Instruct-GGUF:UD-Q8_K_XL -c 8192 -ngl 28
Ollama
ollama run qwen2.5
LM Studio
lms get unsloth/Qwen2.5-VL-7B-Instruct-GGUF # Then load quant UD-Q8_K_XL