Run unsloth/Kimi-K2.7-Code-GGUF locally
unsloth/Kimi-K2.7-Code-GGUF is a very large code-focused language model with 1026.41 billion parameters, built on the deepseek2 architecture. It is released under the other license and has been downloaded 398,867 times.
To run unsloth/Kimi-K2.7-Code-GGUF locally at a 4,096-token context, its quantized versions need between 2.16 GB (F16, lowest quality) and 554.99 GB (Q8_K_XL, highest quality) of memory, weights plus KV cache and a system margin included.
For most users the best balance is F16, needing about 2.16 GB. That means unsloth/Kimi-K2.7-Code-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 | 0.01 | Very low | 0.89 GB | 0.47 GB | 2.16 GB | 450.9 t/s | Fits in VRAM |
| BF16 | 0.01 | Very low | 0.89 GB | 0.47 GB | 2.16 GB | 450.2 t/s | Fits in VRAM |
| F32 | 0.01 | Very low | 1.76 GB | 0.47 GB | 3.03 GB | 227.9 t/s | Fits in VRAM |
| IQ1_M | 2.37 | Very low | 283.04 GB | 0.47 GB | 284.31 GB | — | Insufficient |
| IQ2_XXS | 2.48 | Very low | 296.0 GB | 0.47 GB | 297.27 GB | — | Insufficient |
| IQ2_M | 2.48 | Very low | 296.14 GB | 0.47 GB | 297.41 GB | — | Insufficient |
| Q2_K_XL | 2.65 | Low | 316.15 GB | 0.47 GB | 317.43 GB | — | Insufficient |
| IQ3_S | 3.26 | Low | 390.04 GB | 0.47 GB | 391.32 GB | — | Insufficient |
| Q3_K_M | 3.61 | Fair | 431.78 GB | 0.47 GB | 433.05 GB | — | Insufficient |
| Q3_K_XL | 3.62 | Fair | 432.04 GB | 0.47 GB | 433.32 GB | — | Insufficient |
| IQ4_XS | 3.86 | Fair | 461.08 GB | 0.47 GB | 462.36 GB | — | Insufficient |
| Q4_K_XL | 4.55 | Good | 543.62 GB | 0.47 GB | 544.9 GB | — | Insufficient |
| Q8_K_XL | 4.63 | Good | 553.71 GB | 0.47 GB | 554.99 GB | — | Insufficient |
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/Kimi-K2.7-Code-GGUF?
You need about 2.16 GB of VRAM to run unsloth/Kimi-K2.7-Code-GGUF entirely on the GPU using the F16 quantization (at a 4,096-token context). Smaller quantizations lower the requirement at the cost of quality.
Can I run unsloth/Kimi-K2.7-Code-GGUF on an 8 GB GPU?
Yes. With 8 GB of VRAM you can run unsloth/Kimi-K2.7-Code-GGUF fully on the GPU using F16 (about 2.16 GB).
Can I run unsloth/Kimi-K2.7-Code-GGUF on a 16 GB GPU?
Yes. With 16 GB of VRAM you can run unsloth/Kimi-K2.7-Code-GGUF fully on the GPU using F16 (about 2.16 GB).
Can I run unsloth/Kimi-K2.7-Code-GGUF on a 24 GB GPU?
Yes. With 24 GB of VRAM you can run unsloth/Kimi-K2.7-Code-GGUF fully on the GPU using F16 (about 2.16 GB).
What is the best quantization for unsloth/Kimi-K2.7-Code-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.