Qwen · 9B to 32B
Qwen3 14B
14.77 billion parameters across 40 layers. At four-bit precision the weights alone come to 8.30 GB, before any conversation is loaded. This one is cache-hungry: a 32k window adds 5.0 GB on top, so context length matters more here than the model size suggests.
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 | 14.62 GB | 1.25 GB | 16.78 GB | 99% | Lossless in practice. Use it when the memory is there. |
| Q6_K | 11.28 GB | 1.25 GB | 13.44 GB | 98% | Very close to Q8 for two thirds of the size. |
| Q5_K_M | 9.78 GB | 1.25 GB | 11.95 GB | 96% | The quality-first choice when Q6 will not fit. |
| Q4_K_M | 8.3 GB | 1.25 GB | 10.47 GB | 93% | The default. Best size-to-quality ratio for local use. |
| IQ4_XS | 7.31 GB | 1.25 GB | 9.47 GB | 91% | Importance-matrix 4-bit. Q4_K_S quality, smaller file. |
| Q3_K_M | 6.72 GB | 1.25 GB | 8.89 GB | 86% | Degradation starts to show. A way to fit one size up. |
| IQ3_XXS | 5.26 GB | 1.25 GB | 7.43 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 | 0.63 GB | 9.69 GB | no | yes | yes |
| 8k | 1.25 GB | 10.47 GB | no | yes | yes |
| 16k | 2.5 GB | 12.03 GB | no | no | yes |
| 32k | 5 GB | 15.16 GB | no | no | yes |
The "fits" columns allow for the roughly 0.8 GB Windows keeps for the desktop.
Which hardware runs Qwen3 14B
100 of 118 consumer devices run it at a quantisation worth using, at an 8k context window. Another 11 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 |
|---|---|---|---|---|---|
| GeForce RTX 5090 NVIDIA | 32 GB | Runs great | Q8_0 | 16.78 GB | 92.6 t/s |
| RTX PRO 6000 Blackwell NVIDIA | 96 GB | Runs great | Q8_0 | 16.78 GB | 92.6 t/s |
| GeForce RTX 4090 NVIDIA | 24 GB | Runs great | Q8_0 | 16.78 GB | 52.1 t/s |
| GeForce RTX 3090 Ti NVIDIA | 24 GB | Runs great | Q8_0 | 16.78 GB | 52.1 t/s |
| RTX 6000 Ada Generation NVIDIA | 48 GB | Runs great | Q8_0 | 16.78 GB | 49.6 t/s |
| GeForce RTX 3090 NVIDIA | 24 GB | Runs great | Q8_0 | 16.78 GB | 48.4 t/s |
| GeForce RTX 5090 Laptop NVIDIA | 24 GB | Runs great | Q8_0 | 16.78 GB | 46.3 t/s |
| Radeon RX 7900 XTX AMD | 24 GB | Runs great | Q8_0 | 16.78 GB | 43.6 t/s |
| Apple M3 Ultra 96GB Apple | 96 GB | Runs great | Q8_0 | 16.48 GB | 40.3 t/s |
| Apple M3 Ultra 256GB Apple | 256 GB | Runs great | Q8_0 | 16.48 GB | 40.3 t/s |
| Apple M3 Ultra 512GB Apple | 512 GB | Runs great | Q8_0 | 16.48 GB | 40.3 t/s |
| RTX A6000 NVIDIA | 48 GB | Runs great | Q8_0 | 16.78 GB | 39.7 t/s |
| RTX A5000 NVIDIA | 24 GB | Runs great | Q8_0 | 16.78 GB | 39.7 t/s |
| Apple M1 Ultra 64GB Apple | 64 GB | Runs great | Q8_0 | 16.48 GB | 39.3 t/s |
| Apple M1 Ultra 128GB Apple | 128 GB | Runs great | Q8_0 | 16.48 GB | 39.3 t/s |
| Apple M2 Ultra 64GB Apple | 64 GB | Runs great | Q8_0 | 16.48 GB | 39.3 t/s |
| Apple M2 Ultra 128GB Apple | 128 GB | Runs great | Q8_0 | 16.48 GB | 39.3 t/s |
| Apple M2 Ultra 192GB Apple | 192 GB | Runs great | Q8_0 | 16.48 GB | 39.3 t/s |
| Radeon PRO W7900 AMD | 48 GB | Runs great | Q8_0 | 16.78 GB | 39.2 t/s |
| Radeon RX 7900 XT AMD | 20 GB | Runs great | Q8_0 | 16.78 GB | 36.3 t/s |
| Apple M4 Max 36GB Apple | 36 GB | Runs great | Q8_0 | 16.48 GB | 26.8 t/s |
| Apple M4 Max 48GB Apple | 48 GB | Runs great | Q8_0 | 16.48 GB | 26.8 t/s |
| Apple M4 Max 64GB Apple | 64 GB | Runs great | Q8_0 | 16.48 GB | 26.8 t/s |
| Apple M4 Max 128GB Apple | 128 GB | Runs great | Q8_0 | 16.48 GB | 26.8 t/s |
| Apple M1 Max 32GB Apple | 32 GB | Runs well | Q8_0 | 16.48 GB | 19.7 t/s |
| Apple M1 Max 64GB Apple | 64 GB | Runs well | Q8_0 | 16.48 GB | 19.7 t/s |
| Apple M2 Max 32GB Apple | 32 GB | Runs well | Q8_0 | 16.48 GB | 19.7 t/s |
| Apple M2 Max 64GB Apple | 64 GB | Runs well | Q8_0 | 16.48 GB | 19.7 t/s |
| Apple M2 Max 96GB Apple | 96 GB | Runs well | Q8_0 | 16.48 GB | 19.7 t/s |
| Apple M3 Max 36GB Apple | 36 GB | Runs well | Q8_0 | 16.48 GB | 19.7 t/s |
| Apple M3 Max 48GB Apple | 48 GB | Runs well | Q8_0 | 16.48 GB | 19.7 t/s |
| Apple M3 Max 64GB Apple | 64 GB | Runs well | Q8_0 | 16.48 GB | 19.7 t/s |
| Apple M3 Max 96GB Apple | 96 GB | Runs well | Q8_0 | 16.48 GB | 19.7 t/s |
| Apple M3 Max 128GB Apple | 128 GB | Runs well | Q8_0 | 16.48 GB | 19.7 t/s |
| GeForce RTX 5080 NVIDIA | 16 GB | Runs great | Q6_K | 13.44 GB | 62.8 t/s |
| GeForce RTX 5070 Ti NVIDIA | 16 GB | Runs great | Q6_K | 13.44 GB | 58.6 t/s |
| GeForce RTX 5080 Laptop NVIDIA | 16 GB | Runs great | Q6_K | 13.44 GB | 50.3 t/s |
| GeForce RTX 4080 SUPER NVIDIA | 16 GB | Runs great | Q6_K | 13.44 GB | 48.2 t/s |
| GeForce RTX 4080 NVIDIA | 16 GB | Runs great | Q6_K | 13.44 GB | 46.9 t/s |
| GeForce RTX 4070 Ti SUPER NVIDIA | 16 GB | Runs great | Q6_K | 13.44 GB | 44 t/s |
| GeForce RTX 4090 Laptop NVIDIA | 16 GB | Runs great | Q6_K | 13.44 GB | 37.7 t/s |
| Radeon RX 9070 XT AMD | 16 GB | Runs great | Q6_K | 13.44 GB | 37.1 t/s |
| Radeon RX 9070 AMD | 16 GB | Runs great | Q6_K | 13.44 GB | 37.1 t/s |
| Radeon RX 7800 XT AMD | 16 GB | Runs great | Q6_K | 13.44 GB | 35.9 t/s |
| Radeon RX 7900 GRE AMD | 16 GB | Runs great | Q6_K | 13.44 GB | 33.1 t/s |
| Radeon RX 6900 XT AMD | 16 GB | Runs great | Q6_K | 13.44 GB | 29.4 t/s |
| Radeon RX 6800 AMD | 16 GB | Runs great | Q6_K | 13.44 GB | 29.4 t/s |
| GeForce RTX 5060 Ti 16GB NVIDIA | 16 GB | Runs great | Q6_K | 13.44 GB | 29.3 t/s |
| RTX A4000 NVIDIA | 16 GB | Runs great | Q6_K | 13.44 GB | 29.3 t/s |
| Arc A770 16GB Intel | 16 GB | Runs great | Q6_K | 13.44 GB | 29.1 t/s |
| GeForce RTX 4060 Ti 16GB NVIDIA | 16 GB | Runs well | Q6_K | 13.44 GB | 18.8 t/s |
| Radeon RX 7600 XT AMD | 16 GB | Runs well | Q6_K | 13.44 GB | 16.5 t/s |
| GeForce RTX 3080 Ti NVIDIA | 12 GB | Just fits | Q4_K_M | 10.47 GB | 78.3 t/s |
| GeForce RTX 3080 12GB NVIDIA | 12 GB | Just fits | Q4_K_M | 10.47 GB | 78.3 t/s |
| GeForce RTX 5070 NVIDIA | 12 GB | Just fits | Q4_K_M | 10.47 GB | 57.7 t/s |
| GeForce RTX 4070 Ti NVIDIA | 12 GB | Just fits | Q4_K_M | 10.47 GB | 43.3 t/s |
| GeForce RTX 4070 SUPER NVIDIA | 12 GB | Just fits | Q4_K_M | 10.47 GB | 43.3 t/s |
| GeForce RTX 4070 NVIDIA | 12 GB | Just fits | Q4_K_M | 10.47 GB | 43.3 t/s |
| GeForce RTX 4080 Laptop NVIDIA | 12 GB | Just fits | Q4_K_M | 10.47 GB | 37.1 t/s |
| Radeon RX 7700 XT AMD | 12 GB | Just fits | Q4_K_M | 10.47 GB | 32.6 t/s |
What to buy to run Qwen3 14B
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
GeForce RTX 2060 12GB
$160 used
Q4_K_M · uses 10.47 GB of its 12 GB · about 28.8 tokens/s
Best value
GeForce RTX 3060 12GB
$200 used
Q4_K_M · about 30.9 tokens/s · 360 GB/s
| Device | Price | Memory | Runs it at | Speed | |
|---|---|---|---|---|---|
| GeForce GTX 1080 Ti NVIDIA | $150 used | 11 GB | IQ4_XS | 46.4 t/s | Check price |
| GeForce RTX 2060 12GB NVIDIA | $160 used | 12 GB | Q4_K_M | 28.8 t/s | Check price |
| GeForce RTX 2080 Ti NVIDIA | $250 used | 11 GB | IQ4_XS | 59 t/s | Check price |
| Arc A770 16GB Intel | $280 used | 16 GB | Q6_K | 29.1 t/s | Check price |
| GeForce RTX 3080 12GB NVIDIA | $400 used | 12 GB | Q4_K_M | 78.3 t/s | Check price |
| Radeon RX 7900 XT AMD | $650 | 20 GB | Q8_0 | 36.3 t/s | Check price |
| GeForce RTX 3090 NVIDIA | $700 used | 24 GB | Q8_0 | 48.4 t/s | Check price |
| Ryzen AI Max+ 395 32GB AMD · whole machine | $1,299 | 32 GB | Q5_K_M | 15.1 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.
These numbers also cover
Fine-tunes share their base model's architecture, so they need exactly the same memory. If you are looking for one of these, the figures above apply unchanged.
Qwen/Qwen3-14B-Base
Where it sits in the catalogue
Source: Qwen/Qwen3-14B on Hugging Face. Downloaded 1.8M times in the last month. Published 2025-04-27. Architecture figures are read from the repository's own configuration file, so they move when the model does.
Qwen2.5 14B Instruct
14.77B · 32k context
Qwen1.5 MoE A2.7B
14.32B (2.69B active) · 8k context
Qwen3.5 9B
9.65B · 256k context
Qwen2.5 VL 7B Instruct
8.29B · 125k context
Qwen3 8B
8.19B · 40k context
Qwen1.5 7B
7.72B · 32k context
Qwen2.5 7B Instruct
7.62B · 32k context
Qwen3.5 4B
4.66B · 256k context