Run unsloth/Qwen3-Coder-Next-GGUF locally
unsloth/Qwen3-Coder-Next-GGUF is a very large code-focused language model with 79.67 billion parameters, built on the qwen3next architecture. It is released under the apache-2.0 license and has been downloaded 279,249 times.
To run unsloth/Qwen3-Coder-Next-GGUF locally at a 4,096-token context, its quantized versions need between 18.63 GB (Q1_0, lowest quality) and 149.49 GB (BF16, highest quality) of memory, weights plus KV cache and a system margin included.
For most users the best balance is IQ3_S, needing about 28.64 GB. That means unsloth/Qwen3-Coder-Next-GGUF fits entirely in the VRAM of a 24 GB GPU or larger, running fully on the GPU.
All quantizations
| Quant. | Bits | Quality | Weights | KV | Total | Speed~ | Verdict |
|---|---|---|---|---|---|---|---|
| Q1_0 | 1.9 | Very low | 17.64 GB | 0.19 GB | 18.63 GB | 22.7 t/s | Fits in VRAM |
| IQ1_S | 2.16 | Very low | 20.03 GB | 0.19 GB | 21.02 GB | 20.0 t/s | Fits in VRAM |
| IQ1_M | 2.18 | Very low | 20.21 GB | 0.19 GB | 21.2 GB | 19.8 t/s | Fits in VRAM |
| IQ2_XXS | 2.34 | Very low | 21.71 GB | 0.19 GB | 22.7 GB | 18.4 t/s | Fits in VRAM |
| IQ2_M | 2.51 | Very low | 23.25 GB | 0.19 GB | 24.24 GB | 17.2 t/s | Fits in VRAM |
| Q2_K_XL | 2.69 | Low | 24.92 GB | 0.19 GB | 25.91 GB | 16.0 t/s | Fits in VRAM |
| IQ3_XXS | 2.86 | Low | 26.53 GB | 0.19 GB | 27.51 GB | 15.1 t/s | Fits in VRAM |
| Q2_K | 2.93 | Low | 27.22 GB | 0.19 GB | 28.2 GB | 14.7 t/s | Fits in VRAM |
| Q2_K_L | 2.94 | Low | 27.29 GB | 0.19 GB | 28.27 GB | 14.7 t/s | Fits in VRAM |
| IQ3_S | 2.98 | Low | 27.65 GB | 0.19 GB | 28.64 GB | 14.5 t/s | Fits in VRAM |
| Q3_K_XL | 3.64 | Fair | 33.79 GB | 0.19 GB | 34.78 GB | 1.5 t/s | Offload |
| Q4_0 | 4.55 | Good | 42.22 GB | 0.19 GB | 43.2 GB | 1.2 t/s | Offload |
| GGUF | 4.82 | Good | 44.73 GB | 0.19 GB | 45.72 GB | 1.1 t/s | Offload |
| Q4_K_XL | 4.98 | Good | 46.2 GB | 0.19 GB | 47.19 GB | 1.1 t/s | Offload |
| Q4_1 | 5.03 | Very good | 46.62 GB | 0.19 GB | 47.61 GB | 1.1 t/s | Offload |
| Q5_K_XL | 5.98 | Very good | 55.45 GB | 0.19 GB | 56.44 GB | 0.9 t/s | Offload |
| Q3_K_S | 6.82 | Excellent | 63.26 GB | 0.19 GB | 64.25 GB | — | Insufficient |
| Q6_K_XL | 7.34 | Excellent | 68.1 GB | 0.19 GB | 69.08 GB | — | Insufficient |
| Q3_K_M | 7.46 | Excellent | 69.16 GB | 0.19 GB | 70.15 GB | — | Insufficient |
| IQ4_XS | 8.14 | Excellent | 75.54 GB | 0.19 GB | 76.52 GB | — | Insufficient |
| IQ4_NL | 8.47 | Excellent | 78.57 GB | 0.19 GB | 79.56 GB | — | Insufficient |
| Q8_0 | 8.52 | Excellent | 78.99 GB | 0.19 GB | 79.98 GB | — | Insufficient |
| Q8_K_XL | 8.67 | Excellent | 80.42 GB | 0.19 GB | 81.41 GB | — | Insufficient |
| Q4_K_S | 9.2 | Excellent | 85.32 GB | 0.19 GB | 86.31 GB | — | Insufficient |
| Q4_K_M | 9.82 | Excellent | 91.11 GB | 0.19 GB | 92.1 GB | — | Insufficient |
| Q5_K_S | 11.13 | Excellent | 103.26 GB | 0.19 GB | 104.25 GB | — | Insufficient |
| Q5_K_M | 11.66 | Excellent | 108.12 GB | 0.19 GB | 109.1 GB | — | Insufficient |
| Q6_K | 13.19 | Excellent | 122.38 GB | 0.19 GB | 123.37 GB | — | Insufficient |
| BF16 | 16.01 | Excellent | 148.51 GB | 0.19 GB | 149.49 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/Qwen3-Coder-Next-GGUF?
You need about 22.7 GB of VRAM to run unsloth/Qwen3-Coder-Next-GGUF entirely on the GPU using the IQ2_XXS quantization (at a 4,096-token context). Smaller quantizations lower the requirement at the cost of quality.
Can I run unsloth/Qwen3-Coder-Next-GGUF on an 8 GB GPU?
Partially. unsloth/Qwen3-Coder-Next-GGUF only fits on an 8 GB GPU by offloading part of it to system RAM (with IQ2_XXS), which runs but is slower.
Can I run unsloth/Qwen3-Coder-Next-GGUF on a 16 GB GPU?
Partially. unsloth/Qwen3-Coder-Next-GGUF only fits on a 16 GB GPU by offloading part of it to system RAM (with Q4_1), which runs but is slower.
Can I run unsloth/Qwen3-Coder-Next-GGUF on a 24 GB GPU?
Yes. With 24 GB of VRAM you can run unsloth/Qwen3-Coder-Next-GGUF fully on the GPU using IQ2_XXS (about 22.7 GB).
What is the best quantization for unsloth/Qwen3-Coder-Next-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.