Run TheBloke/Mistral-7B-Instruct-v0.2-GGUF locally
TheBloke/Mistral-7B-Instruct-v0.2-GGUF is a mid-size instruction-tuned chat model with 7.24 billion parameters, built on the llama architecture. It is released under the apache-2.0 license and has been downloaded 155,931 times.
To run TheBloke/Mistral-7B-Instruct-v0.2-GGUF locally at a 4,096-token context, its quantized versions need between 4.17 GB (Q2_K, lowest quality) and 8.47 GB (Q8_0, highest quality) of memory, weights plus KV cache and a system margin included.
For most users the best balance is Q5_0, needing about 5.95 GB. That means TheBloke/Mistral-7B-Instruct-v0.2-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 |
|---|---|---|---|---|---|---|---|
| Q2_K | 3.41 | Fair | 2.87 GB | 0.5 GB | 4.17 GB | 139.3 t/s | Fits in VRAM |
| Q3_K_S | 3.5 | Fair | 2.95 GB | 0.5 GB | 4.25 GB | 135.7 t/s | Fits in VRAM |
| Q3_K_M | 3.89 | Fair | 3.28 GB | 0.5 GB | 4.58 GB | 122.1 t/s | Fits in VRAM |
| Q3_K_L | 4.22 | Good | 3.56 GB | 0.5 GB | 4.86 GB | 112.4 t/s | Fits in VRAM |
| Q4_0 | 4.54 | Good | 3.83 GB | 0.5 GB | 5.13 GB | 104.5 t/s | Fits in VRAM |
| Q4_K_S | 4.57 | Good | 3.86 GB | 0.5 GB | 5.16 GB | 103.7 t/s | Fits in VRAM |
| Q4_K_M | 4.83 | Good | 4.07 GB | 0.5 GB | 5.37 GB | 98.3 t/s | Fits in VRAM |
| Q5_0 | 5.52 | Very good | 4.65 GB | 0.5 GB | 5.95 GB | 85.9 t/s | Fits in VRAM |
| Q5_K_S | 5.52 | Very good | 4.65 GB | 0.5 GB | 5.95 GB | 85.9 t/s | Fits in VRAM |
| Q5_K_M | 5.67 | Very good | 4.78 GB | 0.5 GB | 6.08 GB | 10.5 t/s | Offload |
| Q6_K | 6.56 | Excellent | 5.53 GB | 0.5 GB | 6.83 GB | 9.0 t/s | Offload |
| Q8_0 | 8.5 | Excellent | 7.17 GB | 0.5 GB | 8.47 GB | 7.0 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 TheBloke/Mistral-7B-Instruct-v0.2-GGUF?
You need about 5.95 GB of VRAM to run TheBloke/Mistral-7B-Instruct-v0.2-GGUF entirely on the GPU using the Q5_0 quantization (at a 4,096-token context). Smaller quantizations lower the requirement at the cost of quality.
Can I run TheBloke/Mistral-7B-Instruct-v0.2-GGUF on an 8 GB GPU?
Yes. With 8 GB of VRAM you can run TheBloke/Mistral-7B-Instruct-v0.2-GGUF fully on the GPU using Q6_K (about 6.83 GB).
Can I run TheBloke/Mistral-7B-Instruct-v0.2-GGUF on a 16 GB GPU?
Yes. With 16 GB of VRAM you can run TheBloke/Mistral-7B-Instruct-v0.2-GGUF fully on the GPU using Q8_0 (about 8.47 GB).
Can I run TheBloke/Mistral-7B-Instruct-v0.2-GGUF on a 24 GB GPU?
Yes. With 24 GB of VRAM you can run TheBloke/Mistral-7B-Instruct-v0.2-GGUF fully on the GPU using Q8_0 (about 8.47 GB).
What is the best quantization for TheBloke/Mistral-7B-Instruct-v0.2-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.