unsloth
Qwen3.5 4B
Qwen3.5 4B is a 4.21B dense qwen35 model from unsloth. Configured context length is 262,144 tokens. License: apache-2.0. Quant sizes are the GGUF file bytes published on Hugging Face when a file exists.
Parameters4.2B
Context256K
Attentionhybrid
Licenseapache-2.0
32 layers · 4 KV heads · head dim 256 · 24 linear-attention layers · GGUF repo · config
Quant file sizes
| Quant | File size | Source |
|---|---|---|
| BF16 | 7.85 GB | Hugging Face file bytes |
| UD-Q8_K_XL | 5.54 GB | Hugging Face file bytes |
| Q8_0 | 4.17 GB | Hugging Face file bytes |
| UD-Q6_K_XL | 3.86 GB | Hugging Face file bytes |
| Q6_K | 3.28 GB | Hugging Face file bytes |
| UD-Q5_K_XL | 3.03 GB | Hugging Face file bytes |
| Q5_K_M | 2.93 GB | Hugging Face file bytes |
| Q5_K_S | 2.82 GB | Hugging Face file bytes |
| UD-Q4_K_XL | 2.71 GB | Hugging Face file bytes |
| Q4_K_M | 2.55 GB | Hugging Face file bytes |
| Q4_K_S | 2.41 GB | Hugging Face file bytes |
| IQ4_NL | 2.40 GB | Hugging Face file bytes |
| IQ4_XS | 2.31 GB | Hugging Face file bytes |
| Q4_1 | 2.59 GB | Hugging Face file bytes |
| Q4_0 | 2.41 GB | Hugging Face file bytes |
| UD-Q3_K_XL | 2.27 GB | Hugging Face file bytes |
| Q3_K_M | 2.14 GB | Hugging Face file bytes |
| Q3_K_S | 1.96 GB | Hugging Face file bytes |
| UD-IQ3_XXS | 1.82 GB | Hugging Face file bytes |
| UD-Q2_K_XL | 1.81 GB | Hugging Face file bytes |
| UD-IQ2_XXS | 1.42 GB | Hugging Face file bytes |
On common machines
| Device | Verdict | Quant | Speed |
|---|---|---|---|
| GeForce RTX 4060 Ti 8GB | Tight fit | UD-Q8_K_XL | 25–35 |
| GeForce RTX 3060 12GB | Runs | BF16 | 23–31 |
| GeForce RTX 4060 Ti 16GB | Runs great | BF16 | 18–25 |
| GeForce RTX 4090 | Runs great | BF16 | 64–87 |
| GeForce RTX 5090 | Runs great | BF16 | 113–155 |
| Apple M4 Pro | Runs great | BF16 | 17–24 |
| Apple M5 | Runs | BF16 | 9.7–13 |
| CPU only · DDR5-5600 | Runs on CPU | BF16 | 5.7–7.7 |
Smallest devices that fit it fully
- GeForce RTX 3050 6GB — Tight fit, Q8_0, 19–27 tok/s
- GeForce RTX 2060 6GB — Tight fit, Q8_0, 39–53 tok/s
- GeForce GTX 1660 Super — Tight fit, Q8_0, 39–53 tok/s
- GeForce GTX 1660 Ti — Tight fit, Q8_0, 33–45 tok/s
- Intel Arc A380 — Tight fit, Q8_0, 22–29 tok/s
- GeForce RTX 5060 Ti 8GB — Tight fit, UD-Q8_K_XL, 40–54 tok/s
Run it
Commands for the GeForce RTX 4060 Ti 16GB estimate (BF16).
llama.cpp
llama-server -hf unsloth/Qwen3.5-4B-GGUF:BF16 -c 8192 -ngl 32
Ollama
ollama run qwen3.5
LM Studio
lms get unsloth/Qwen3.5-4B-GGUF # Then load quant BF16