unsloth
Qwen3 8B
Qwen3 8B is a 8.19B dense qwen3 model from unsloth. Configured context length is 40,960 tokens. License: apache-2.0. Quant sizes are the GGUF file bytes published on Hugging Face when a file exists.
Parameters8.2B
Context40K
Attentiongqa
Licenseapache-2.0
36 layers · 8 KV heads · head dim 128 · GGUF repo · config
Quant file sizes
| Quant | File size | Source |
|---|---|---|
| BF16 | 15.3 GB | Hugging Face file bytes |
| UD-Q8_K_XL | 10.1 GB | Hugging Face file bytes |
| Q8_0 | 8.11 GB | Hugging Face file bytes |
| UD-Q6_K_XL | 6.98 GB | Hugging Face file bytes |
| Q6_K | 6.26 GB | Hugging Face file bytes |
| UD-Q5_K_XL | 5.47 GB | Hugging Face file bytes |
| Q5_K_M | 5.45 GB | Hugging Face file bytes |
| Q5_K_S | 5.33 GB | Hugging Face file bytes |
| UD-Q4_K_XL | 4.78 GB | Hugging Face file bytes |
| Q4_K_M | 4.68 GB | Hugging Face file bytes |
| Q4_K_S | 4.47 GB | Hugging Face file bytes |
| IQ4_NL | 4.46 GB | Hugging Face file bytes |
| IQ4_XS | 4.27 GB | Hugging Face file bytes |
| Q4_1 | 4.89 GB | Hugging Face file bytes |
| UD-Q3_K_XL | 4.01 GB | Hugging Face file bytes |
| Q3_K_M | 3.84 GB | Hugging Face file bytes |
| Q3_K_S | 3.51 GB | Hugging Face file bytes |
| UD-IQ3_XXS | 3.18 GB | Hugging Face file bytes |
| UD-Q2_K_XL | 3.26 GB | Hugging Face file bytes |
| Q2_K | 6.25 GB | Hugging Face file bytes |
| UD-IQ2_XXS | 2.43 GB | Hugging Face file bytes |
| UD-IQ1_M | 2.23 GB | Hugging Face file bytes |
| UD-IQ1_S | 2.12 GB | Hugging Face file bytes |
On common machines
| Device | Verdict | Quant | Speed |
|---|---|---|---|
| GeForce RTX 4060 Ti 8GB | Tight fit | UD-Q5_K_XL | 22–30 |
| GeForce RTX 3060 12GB | Tight fit | Q8_0 | 20–27 |
| GeForce RTX 4060 Ti 16GB | Runs | UD-Q8_K_XL | 13–18 |
| GeForce RTX 4090 | Runs | BF16 | 32–43 |
| GeForce RTX 5090 | Runs great | BF16 | 56–76 |
| Apple M4 Pro | Tight fit | BF16 | 8.5–12 |
| Apple M5 | Tight fit | Q8_0 | 8.5–12 |
| CPU only · DDR5-5600 | Runs on CPU | BF16 | 2.8–3.8 |
Smallest devices that fit it fully
- GeForce RTX 3050 6GB — Tight fit, UD-Q3_K_XL, 19–26 tok/s
- GeForce RTX 2060 6GB — Tight fit, UD-Q3_K_XL, 38–51 tok/s
- GeForce GTX 1660 Super — Tight fit, UD-Q3_K_XL, 38–51 tok/s
- GeForce GTX 1660 Ti — Tight fit, UD-Q3_K_XL, 32–44 tok/s
- Intel Arc A380 — Tight fit, UD-Q3_K_XL, 21–28 tok/s
- GeForce RTX 5060 Ti 8GB — Tight fit, UD-Q5_K_XL, 35–47 tok/s
Run it
Commands for the GeForce RTX 4060 Ti 16GB estimate (UD-Q8_K_XL).
llama.cpp
llama-server -hf unsloth/Qwen3-8B-GGUF:UD-Q8_K_XL -c 8192 -ngl 36
Ollama
ollama run qwen3
LM Studio
lms get unsloth/Qwen3-8B-GGUF # Then load quant UD-Q8_K_XL