Run DavidAU/Qwen3.6-40B-Claude-4.6-Opus-Deckard-Heretic-Uncensored-Thinking-NEO-CODE-Di-IMatrix-MAX-GGUF locally
DavidAU/Qwen3.6-40B-Claude-4.6-Opus-Deckard-Heretic-Uncensored-Thinking-NEO-CODE-Di-IMatrix-MAX-GGUF is a very large code-focused language model with 39.07 billion parameters, built on the qwen35 architecture. It is released under the apache-2.0 license and has been downloaded 544,924 times.
To run DavidAU/Qwen3.6-40B-Claude-4.6-Opus-Deckard-Heretic-Uncensored-Thinking-NEO-CODE-Di-IMatrix-MAX-GGUF locally at a 4,096-token context, its quantized versions need between 2.14 GB (F16, lowest quality) and 41.18 GB (Q8_0, highest quality) of memory, weights plus KV cache and a system margin included.
For most users the best balance is F32, needing about 2.99 GB. That means DavidAU/Qwen3.6-40B-Claude-4.6-Opus-Deckard-Heretic-Uncensored-Thinking-NEO-CODE-Di-IMatrix-MAX-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 | 0.19 | Very low | 0.86 GB | 0.47 GB | 2.14 GB | 463.0 t/s | Fits in VRAM |
| BF16 | 0.19 | Very low | 0.87 GB | 0.47 GB | 2.14 GB | 461.3 t/s | Fits in VRAM |
| F32 | 0.38 | Very low | 1.72 GB | 0.47 GB | 2.99 GB | 233.0 t/s | Fits in VRAM |
| IQ2_M | 3.03 | Low | 13.76 GB | 0.47 GB | 15.04 GB | 3.6 t/s | Offload |
| IQ3_M | 3.76 | Fair | 17.12 GB | 0.47 GB | 18.39 GB | 2.9 t/s | Offload |
| IQ4_XS | 4.52 | Good | 20.56 GB | 0.47 GB | 21.84 GB | 2.4 t/s | Offload |
| Q4_K_S | 4.67 | Good | 21.24 GB | 0.47 GB | 22.51 GB | 2.4 t/s | Offload |
| IQ4_NL | 4.74 | Good | 21.55 GB | 0.47 GB | 22.82 GB | 2.3 t/s | Offload |
| Q4_K_M | 4.97 | Good | 22.59 GB | 0.47 GB | 23.86 GB | 2.2 t/s | Offload |
| Q5_K_S | 5.6 | Very good | 25.49 GB | 0.47 GB | 26.76 GB | — | Insufficient |
| Q5_K_M | 5.77 | Very good | 26.26 GB | 0.47 GB | 27.53 GB | — | Insufficient |
| Q6_K | 6.63 | Excellent | 30.17 GB | 0.47 GB | 31.44 GB | — | Insufficient |
| Q8_0 | 8.77 | Excellent | 39.9 GB | 0.47 GB | 41.18 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 DavidAU/Qwen3.6-40B-Claude-4.6-Opus-Deckard-Heretic-Uncensored-Thinking-NEO-CODE-Di-IMatrix-MAX-GGUF?
You need about 2.99 GB of VRAM to run DavidAU/Qwen3.6-40B-Claude-4.6-Opus-Deckard-Heretic-Uncensored-Thinking-NEO-CODE-Di-IMatrix-MAX-GGUF entirely on the GPU using the F32 quantization (at a 4,096-token context). Smaller quantizations lower the requirement at the cost of quality.
Can I run DavidAU/Qwen3.6-40B-Claude-4.6-Opus-Deckard-Heretic-Uncensored-Thinking-NEO-CODE-Di-IMatrix-MAX-GGUF on an 8 GB GPU?
Yes. With 8 GB of VRAM you can run DavidAU/Qwen3.6-40B-Claude-4.6-Opus-Deckard-Heretic-Uncensored-Thinking-NEO-CODE-Di-IMatrix-MAX-GGUF fully on the GPU using F32 (about 2.99 GB).
Can I run DavidAU/Qwen3.6-40B-Claude-4.6-Opus-Deckard-Heretic-Uncensored-Thinking-NEO-CODE-Di-IMatrix-MAX-GGUF on a 16 GB GPU?
Yes. With 16 GB of VRAM you can run DavidAU/Qwen3.6-40B-Claude-4.6-Opus-Deckard-Heretic-Uncensored-Thinking-NEO-CODE-Di-IMatrix-MAX-GGUF fully on the GPU using IQ2_M (about 15.04 GB).
Can I run DavidAU/Qwen3.6-40B-Claude-4.6-Opus-Deckard-Heretic-Uncensored-Thinking-NEO-CODE-Di-IMatrix-MAX-GGUF on a 24 GB GPU?
Yes. With 24 GB of VRAM you can run DavidAU/Qwen3.6-40B-Claude-4.6-Opus-Deckard-Heretic-Uncensored-Thinking-NEO-CODE-Di-IMatrix-MAX-GGUF fully on the GPU using Q4_K_M (about 23.86 GB).
What is the best quantization for DavidAU/Qwen3.6-40B-Claude-4.6-Opus-Deckard-Heretic-Uncensored-Thinking-NEO-CODE-Di-IMatrix-MAX-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.