Run unsloth/Qwen2.5-VL-7B-Instruct-GGUF locally
unsloth/Qwen2.5-VL-7B-Instruct-GGUF is a mid-size instruction-tuned chat model with 7.62 billion parameters, built on the qwen2vl architecture. It is released under the apache-2.0 license and has been downloaded 384,678 times.
To run unsloth/Qwen2.5-VL-7B-Instruct-GGUF locally at a 4,096-token context, its quantized versions need between 2.28 GB (F16, lowest quality) and 16.47 GB (BF16, highest quality) of memory, weights plus KV cache and a system margin included.
For most users the best balance is Q6_K_XL, needing about 7.5 GB. That means unsloth/Qwen2.5-VL-7B-Instruct-GGUF fits entirely in the VRAM of a 6 GB GPU or larger, running fully on the GPU.
All quantizations
| Quant. | Bits | Quality | Weights | KV | Total | Speed~ | Verdict |
|---|---|---|---|---|---|---|---|
| F16 | 1.42 | Very low | 1.26 GB | 0.22 GB | 2.28 GB | 317.2 t/s | Fits in VRAM |
| IQ1_S | 2.18 | Very low | 1.93 GB | 0.22 GB | 2.95 GB | 207.1 t/s | Fits in VRAM |
| IQ1_M | 2.31 | Very low | 2.05 GB | 0.22 GB | 3.07 GB | 195.2 t/s | Fits in VRAM |
| IQ2_XXS | 2.52 | Very low | 2.23 GB | 0.22 GB | 3.25 GB | 179.1 t/s | Fits in VRAM |
| F32 | 2.84 | Low | 2.52 GB | 0.22 GB | 3.54 GB | 158.9 t/s | Fits in VRAM |
| IQ2_M | 3.0 | Low | 2.66 GB | 0.22 GB | 3.68 GB | 150.4 t/s | Fits in VRAM |
| Q2_K | 3.17 | Low | 2.81 GB | 0.22 GB | 3.83 GB | 142.4 t/s | Fits in VRAM |
| Q2_K_L | 3.3 | Fair | 2.93 GB | 0.22 GB | 3.95 GB | 136.6 t/s | Fits in VRAM |
| IQ3_XXS | 3.33 | Fair | 2.95 GB | 0.22 GB | 3.97 GB | 135.6 t/s | Fits in VRAM |
| Q2_K_XL | 3.41 | Fair | 3.02 GB | 0.22 GB | 4.04 GB | 132.4 t/s | Fits in VRAM |
| Q3_K_S | 3.67 | Fair | 3.25 GB | 0.22 GB | 4.27 GB | 123.0 t/s | Fits in VRAM |
| Q3_K_M | 4.0 | Fair | 3.55 GB | 0.22 GB | 4.57 GB | 112.8 t/s | Fits in VRAM |
| Q3_K_XL | 4.2 | Good | 3.73 GB | 0.22 GB | 4.75 GB | 107.3 t/s | Fits in VRAM |
| IQ4_XS | 4.45 | Good | 3.94 GB | 0.22 GB | 4.96 GB | 101.4 t/s | Fits in VRAM |
| IQ4_NL | 4.66 | Good | 4.13 GB | 0.22 GB | 5.15 GB | 96.8 t/s | Fits in VRAM |
| Q4_0 | 4.67 | Good | 4.14 GB | 0.22 GB | 5.16 GB | 96.6 t/s | Fits in VRAM |
| Q4_K_S | 4.68 | Good | 4.15 GB | 0.22 GB | 5.17 GB | 96.3 t/s | Fits in VRAM |
| Q4_K_M | 4.92 | Good | 4.36 GB | 0.22 GB | 5.38 GB | 91.7 t/s | Fits in VRAM |
| Q4_K_XL | 5.03 | Very good | 4.46 GB | 0.22 GB | 5.48 GB | 89.8 t/s | Fits in VRAM |
| Q4_1 | 5.12 | Very good | 4.54 GB | 0.22 GB | 5.56 GB | 88.1 t/s | Fits in VRAM |
| Q5_K_S | 5.58 | Very good | 4.95 GB | 0.22 GB | 5.97 GB | 80.8 t/s | Fits in VRAM |
| Q5_K_M | 5.72 | Very good | 5.07 GB | 0.22 GB | 6.09 GB | 78.9 t/s | Fits in VRAM |
| Q5_K_XL | 5.74 | Very good | 5.09 GB | 0.22 GB | 6.11 GB | 78.6 t/s | Fits in VRAM |
| Q6_K | 6.57 | Excellent | 5.82 GB | 0.22 GB | 6.84 GB | 68.7 t/s | Fits in VRAM |
| Q6_K_XL | 7.31 | Excellent | 6.48 GB | 0.22 GB | 7.5 GB | 61.7 t/s | Fits in VRAM |
| Q8_0 | 8.51 | Excellent | 7.54 GB | 0.22 GB | 8.56 GB | 6.6 t/s | Offload |
| Q8_K_XL | 10.71 | Excellent | 9.49 GB | 0.22 GB | 10.51 GB | 5.3 t/s | Offload |
| BF16 | 17.43 | Excellent | 15.45 GB | 0.22 GB | 16.47 GB | 3.2 t/s | Offload |
KV cache computed from the model's exact architecture. Speed is a rough estimate bounded by memory bandwidth.
Frequently asked questions
How much VRAM do you need to run unsloth/Qwen2.5-VL-7B-Instruct-GGUF?
You need about 5.97 GB of VRAM to run unsloth/Qwen2.5-VL-7B-Instruct-GGUF entirely on the GPU using the Q5_K_S quantization (at a 4,096-token context). Smaller quantizations lower the requirement at the cost of quality.
Can I run unsloth/Qwen2.5-VL-7B-Instruct-GGUF on an 8 GB GPU?
Yes. With 8 GB of VRAM you can run unsloth/Qwen2.5-VL-7B-Instruct-GGUF fully on the GPU using Q6_K_XL (about 7.5 GB).
Can I run unsloth/Qwen2.5-VL-7B-Instruct-GGUF on a 16 GB GPU?
Yes. With 16 GB of VRAM you can run unsloth/Qwen2.5-VL-7B-Instruct-GGUF fully on the GPU using Q8_K_XL (about 10.51 GB).
Can I run unsloth/Qwen2.5-VL-7B-Instruct-GGUF on a 24 GB GPU?
Yes. With 24 GB of VRAM you can run unsloth/Qwen2.5-VL-7B-Instruct-GGUF fully on the GPU using BF16 (about 16.47 GB).
What is the best quantization for unsloth/Qwen2.5-VL-7B-Instruct-GGUF?
If memory allows, higher bits-per-weight means better quality. A common sweet spot is a Q4_K_M or Q5_K_M quantization, which keeps most of the quality while roughly halving the memory versus 8-bit. Pick the highest quantization that still fits in your VRAM.