Qwen · 32B to 80B · Mixture of experts
Qwen3 Coder Next
79.67 billion parameters in total, but only 3.19 billion are read for each token. That split is the whole point of the design: it costs the memory of a large model and the speed of a small one. Grouped-query attention keeps the cache small, so long contexts cost less here than on models of the same size.
Memory needed, by quantisation
At an 8k context window. Quality is our rough ranking of how much the compression costs you: anything at or above Q5 is hard to tell apart from the original in normal use.
| Quantisation | File size | KV cache | Total VRAM | Quality | What it costs you |
|---|---|---|---|---|---|
| Q8_0 | 78.84 GB | 0.75 GB | 80.31 GB | 99% | Lossless in practice. Use it when the memory is there. |
| Q6_K | 60.84 GB | 0.75 GB | 62.32 GB | 98% | Very close to Q8 for two thirds of the size. |
| Q5_K_M | 52.77 GB | 0.75 GB | 54.25 GB | 96% | The quality-first choice when Q6 will not fit. |
| Q4_K_M | 44.8 GB | 0.75 GB | 46.27 GB | 93% | The default. Best size-to-quality ratio for local use. |
| IQ4_XS | 39.42 GB | 0.75 GB | 40.89 GB | 91% | Importance-matrix 4-bit. Q4_K_S quality, smaller file. |
| Q3_K_M | 36.26 GB | 0.75 GB | 37.74 GB | 86% | Degradation starts to show. A way to fit one size up. |
| IQ3_XXS | 28.38 GB | 0.75 GB | 29.86 GB | 79% | Aggressive. Only worth it on very large models. |
What a longer conversation costs
Same model at Q4_K_M, only the context window changes. The cache grows in a straight line with every token in the window.
| Context | KV cache | Total VRAM | Fits in 8 GB | Fits in 12 GB | Fits in 24 GB |
|---|---|---|---|---|---|
| 4k | 0.38 GB | 45.84 GB | no | no | no |
| 8k | 0.75 GB | 46.27 GB | no | no | no |
| 16k | 1.5 GB | 47.15 GB | no | no | no |
| 32k | 3 GB | 48.9 GB | no | no | no |
| 64k | 6 GB | 52.4 GB | no | no | no |
| 128k | 12 GB | 59.4 GB | no | no | no |
The "fits" columns allow for the roughly 0.8 GB Windows keeps for the desktop.
Which hardware runs Qwen3 Coder Next
26 of 118 consumer devices run it at a quantisation worth using, at an 8k context window. Another 23 can load it only by compressing the weights far enough to damage the model, marked with a warning below.
| Device | Memory | Verdict | Quantisation | Used | Speed |
|---|---|---|---|---|---|
| RTX PRO 6000 Blackwell NVIDIA | 96 GB | Runs great | Q8_0 | 80.31 GB | 376.1 t/s |
| Apple M3 Ultra 256GB Apple | 256 GB | Runs great | Q8_0 | 80.01 GB | 163.5 t/s |
| Apple M3 Ultra 512GB Apple | 512 GB | Runs great | Q8_0 | 80.01 GB | 163.5 t/s |
| Apple M1 Ultra 128GB Apple | 128 GB | Runs great | Q8_0 | 80.01 GB | 159.7 t/s |
| Apple M2 Ultra 128GB Apple | 128 GB | Runs great | Q8_0 | 80.01 GB | 159.7 t/s |
| Apple M2 Ultra 192GB Apple | 192 GB | Runs great | Q8_0 | 80.01 GB | 159.7 t/s |
| Apple M4 Max 128GB Apple | 128 GB | Runs great | Q8_0 | 80.01 GB | 109 t/s |
| Apple M3 Max 128GB Apple | 128 GB | Runs great | Q8_0 | 80.01 GB | 79.9 t/s |
| NVIDIA DGX Spark 128GB NVIDIA | 128 GB | Runs great | Q8_0 | 80.31 GB | 57.3 t/s |
| Ryzen AI Max+ 395 128GB AMD | 128 GB | Runs great | Q8_0 | 80.31 GB | 42.6 t/s |
| Apple M3 Ultra 96GB Apple | 96 GB | Runs great | Q6_K | 62.02 GB | 200.5 t/s |
| Apple M2 Max 96GB Apple | 96 GB | Runs great | Q6_K | 62.02 GB | 97.9 t/s |
| Apple M3 Max 96GB Apple | 96 GB | Runs great | Q6_K | 62.02 GB | 97.9 t/s |
| Ryzen AI Max+ 395 96GB AMD | 96 GB | Runs great | Q6_K | 62.32 GB | 52.2 t/s |
| RTX 6000 Ada Generation NVIDIA | 48 GB | Just fits | Q4_K_M | 46.27 GB | 309.5 t/s |
| RTX A6000 NVIDIA | 48 GB | Just fits | Q4_K_M | 46.27 GB | 247.6 t/s |
| Apple M1 Ultra 64GB Apple | 64 GB | Runs great | Q4_K_M | 45.97 GB | 245.3 t/s |
| Apple M2 Ultra 64GB Apple | 64 GB | Runs great | Q4_K_M | 45.97 GB | 245.3 t/s |
| Radeon PRO W7900 AMD | 48 GB | Just fits | Q4_K_M | 46.27 GB | 244.6 t/s |
| Apple M4 Max 64GB Apple | 64 GB | Runs great | Q4_K_M | 45.97 GB | 167.4 t/s |
| Apple M1 Max 64GB Apple | 64 GB | Runs great | Q4_K_M | 45.97 GB | 122.7 t/s |
| Apple M2 Max 64GB Apple | 64 GB | Runs great | Q4_K_M | 45.97 GB | 122.7 t/s |
| Apple M3 Max 64GB Apple | 64 GB | Runs great | Q4_K_M | 45.97 GB | 122.7 t/s |
| Apple M4 Pro 64GB Apple | 64 GB | Runs great | Q4_K_M | 45.97 GB | 83.7 t/s |
| Jetson AGX Orin 64GB NVIDIA | 64 GB | Runs great | Q4_K_M | 46.27 GB | 66 t/s |
| Ryzen AI Max+ 395 64GB AMD | 64 GB | Runs great | Q4_K_M | 46.27 GB | 65.4 t/s |
| GeForce RTX 5090 NVIDIA | 32 GB | Runs great | IQ3_XXS ! | 29.86 GB | 779 t/s |
| Apple M4 Max 48GB Apple | 48 GB | Runs great | IQ3_XXS ! | 29.56 GB | 225.8 t/s |
| Apple M3 Max 48GB Apple | 48 GB | Runs great | IQ3_XXS ! | 29.56 GB | 165.4 t/s |
| Apple M4 Pro 48GB Apple | 48 GB | Runs great | IQ3_XXS ! | 29.56 GB | 112.9 t/s |
| GeForce RTX 4090 NVIDIA | 24 GB | Runs great | IQ2_XXS ! | 20.58 GB | 545.6 t/s |
| GeForce RTX 3090 Ti NVIDIA | 24 GB | Runs great | IQ2_XXS ! | 20.58 GB | 545.6 t/s |
| GeForce RTX 3090 NVIDIA | 24 GB | Runs great | IQ2_XXS ! | 20.58 GB | 506.6 t/s |
| GeForce RTX 5090 Laptop NVIDIA | 24 GB | Runs great | IQ2_XXS ! | 20.58 GB | 485 t/s |
| Radeon RX 7900 XTX AMD | 24 GB | Runs great | IQ2_XXS ! | 20.58 GB | 456.2 t/s |
| RTX A5000 NVIDIA | 24 GB | Runs great | IQ2_XXS ! | 20.58 GB | 415.7 t/s |
| Apple M4 Max 36GB Apple | 36 GB | Runs great | IQ2_XXS ! | 20.28 GB | 281.1 t/s |
| Apple M1 Max 32GB Apple | 32 GB | Runs great | IQ2_XXS ! | 20.28 GB | 205.9 t/s |
| Apple M2 Max 32GB Apple | 32 GB | Runs great | IQ2_XXS ! | 20.28 GB | 205.9 t/s |
| Apple M3 Max 36GB Apple | 36 GB | Runs great | IQ2_XXS ! | 20.28 GB | 205.9 t/s |
| Jetson AGX Orin 32GB NVIDIA | 32 GB | Runs great | IQ2_XXS ! | 20.58 GB | 110.8 t/s |
| Ryzen AI Max+ 395 32GB AMD | 32 GB | Runs great | IQ2_XXS ! | 20.58 GB | 109.8 t/s |
| Apple M1 Pro 32GB Apple | 32 GB | Runs great | IQ2_XXS ! | 20.28 GB | 103 t/s |
| Apple M2 Pro 32GB Apple | 32 GB | Runs great | IQ2_XXS ! | 20.28 GB | 103 t/s |
| Apple M5 32GB Apple | 32 GB | Runs great | IQ2_XXS ! | 20.28 GB | 78.8 t/s |
| Apple M3 Pro 36GB Apple | 36 GB | Runs great | IQ2_XXS ! | 20.28 GB | 77.2 t/s |
| Apple M4 32GB Apple | 32 GB | Runs great | IQ2_XXS ! | 20.28 GB | 61.8 t/s |
| Intel Core Ultra 9 288V 32GB Intel | 32 GB | Runs great | IQ2_XXS ! | 20.58 GB | 58.3 t/s |
| Ryzen AI 9 HX 370 32GB AMD | 32 GB | Runs great | IQ2_XXS ! | 20.58 GB | 54.9 t/s |
| Radeon RX 7900 XT AMD | 20 GB | Partial offload 19/48 | — | 46.27 GB | 19.9 t/s |
| GeForce RTX 5080 NVIDIA | 16 GB | Partial offload 15/48 | — | 46.27 GB | 17.8 t/s |
| GeForce RTX 5070 Ti NVIDIA | 16 GB | Partial offload 15/48 | — | 46.27 GB | 17.7 t/s |
| GeForce RTX 4080 SUPER NVIDIA | 16 GB | Partial offload 15/48 | — | 46.27 GB | 17.7 t/s |
| GeForce RTX 5080 Laptop NVIDIA | 16 GB | Partial offload 15/48 | — | 46.27 GB | 17.7 t/s |
| GeForce RTX 4080 NVIDIA | 16 GB | Partial offload 15/48 | — | 46.27 GB | 17.6 t/s |
| GeForce RTX 4070 Ti SUPER NVIDIA | 16 GB | Partial offload 15/48 | — | 46.27 GB | 17.6 t/s |
| GeForce RTX 4090 Laptop NVIDIA | 16 GB | Partial offload 15/48 | — | 46.27 GB | 17.5 t/s |
| Radeon RX 9070 XT AMD | 16 GB | Partial offload 15/48 | — | 46.27 GB | 17.5 t/s |
| Radeon RX 9070 AMD | 16 GB | Partial offload 15/48 | — | 46.27 GB | 17.5 t/s |
| Radeon RX 7900 GRE AMD | 16 GB | Partial offload 15/48 | — | 46.27 GB | 17.5 t/s |
What to buy to run Qwen3 Coder Next
The cheapest hardware that runs it at a quantisation worth using and a speed you would not resent, at an 8k context window.
Cheapest that works
Ryzen AI Max+ 395 64GB
$1,699 whole machine
Q4_K_M · uses 46.27 GB of its 64 GB · about 65.4 tokens/s
Best value
Ryzen AI Max+ 395 96GB
$1,999 whole machine
Q6_K · about 52.2 tokens/s · 256 GB/s
| Device | Price | Memory | Runs it at | Speed | |
|---|---|---|---|---|---|
| Ryzen AI Max+ 395 64GB AMD · whole machine | $1,699 | 64 GB | Q4_K_M | 65.4 t/s | Check price |
| Apple M1 Max 64GB Apple · whole machine | $1,900 | 64 GB | Q4_K_M | 122.7 t/s | Check price |
| Ryzen AI Max+ 395 96GB AMD · whole machine | $1,999 | 96 GB | Q6_K | 52.2 t/s | Check price |
| Ryzen AI Max+ 395 128GB AMD · whole machine | $2,199 | 128 GB | Q8_0 | 42.6 t/s | Check price |
| Apple M1 Ultra 64GB Apple · whole machine | $2,400 | 64 GB | Q4_K_M | 245.3 t/s | Check price |
| RTX A6000 NVIDIA | $3,500 used | 48 GB | Q4_K_M | 247.6 t/s | Check price |
| Apple M3 Ultra 256GB Apple · whole machine | $5,599 | 256 GB | Q8_0 | 163.5 t/s | Check price |
| RTX 6000 Ada Generation NVIDIA | $6,800 | 48 GB | Q4_K_M | 309.5 t/s | Check price |
Indicative prices reviewed 2026-09-01; used prices are marketplace typical. Options that cost more than a cheaper one with no more memory and no more speed are hidden. Price links are Amazon affiliate links. Change the standard, the context window or the budget in the buying tool.
Where it sits in the catalogue
Source: Qwen/Qwen3-Coder-Next on Hugging Face. Downloaded 450k times in the last month. Published 2026-01-30. Architecture figures are read from the repository's own configuration file, so they move when the model does.
Qwen3 Next 80B A3B Instruct
81.32B (3.19B active) · 256k context
Qwen2.5 VL 72B Instruct
73.41B · 125k context
Qwen2.5 72B Instruct
72.71B · 32k context
Qwen3.5 35B A3B
35.95B (2.9B active) · 256k context
Qwen3.5 122B A10B
125.09B (8.17B active) · 256k context
Qwen3 32B
32.76B · 40k context
Qwen2.5 32B Instruct
32.76B · 32k context
Qwen3 30B A3B
30.53B (3.34B active) · 40k context