unsloth
Qwen3.5 397B A17B
Qwen3.5 397B A17B is a 396.3B mixture-of-experts qwen35moe model from unsloth, with about 17.0B parameters active per token. 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.
Parameters396B / 17.0B active
Context256K
Attentionhybrid
Licenseapache-2.0
60 layers · 2 KV heads · head dim 256 · 45 linear-attention layers · GGUF repo · config
Quant file sizes
| Quant | File size | Source |
|---|---|---|
| BF16 | 738.5 GB | Hugging Face file bytes |
| UD-Q8_K_XL | 398.4 GB | Hugging Face file bytes |
| Q8_0 | 392.6 GB | Hugging Face file bytes |
| UD-Q6_K_XL | 337.4 GB | Hugging Face file bytes |
| UD-Q6_K | 304.2 GB | Hugging Face file bytes |
| Q6_K | 304.2 GB | Hugging Face file bytes |
| UD-Q5_K_XL | 274.6 GB | Hugging Face file bytes |
| UD-Q5_K_M | 273.5 GB | Hugging Face file bytes |
| Q5_K_M | 273.5 GB | Hugging Face file bytes |
| UD-Q5_K_S | 257.6 GB | Hugging Face file bytes |
| Q5_K_S | 257.6 GB | Hugging Face file bytes |
| UD-Q4_K_XL | 228.4 GB | Hugging Face file bytes |
| UD-Q4_K_M | 227.3 GB | Hugging Face file bytes |
| Q4_K_M | 227.3 GB | Hugging Face file bytes |
| UD-Q4_K_S | 212.3 GB | Hugging Face file bytes |
| Q4_K_S | 212.3 GB | Hugging Face file bytes |
| UD-IQ4_XS | 176.7 GB | Hugging Face file bytes |
| IQ4_NL | 180.5 GB | Hugging Face file bytes |
| UD-Q3_K_XL | 166.5 GB | Hugging Face file bytes |
| Q3_K_M | 330.5 GB | Hugging Face file bytes |
| UD-IQ3_S | 136.3 GB | Hugging Face file bytes |
| Q3_K_S | 306.1 GB | Hugging Face file bytes |
| UD-IQ3_XXS | 130.7 GB | Hugging Face file bytes |
| UD-IQ2_XXS | 107.0 GB | Hugging Face file bytes |
| UD-IQ1_M | 99.5 GB | Hugging Face file bytes |
On common machines
| Device | Verdict | Quant | Speed |
|---|---|---|---|
| GeForce RTX 4060 Ti 8GB | Won't fit | UD-IQ1_M | — |
| GeForce RTX 3060 12GB | Won't fit | UD-IQ1_M | — |
| GeForce RTX 4060 Ti 16GB | Won't fit | UD-IQ1_M | — |
| GeForce RTX 4090 | Won't fit | UD-IQ1_M | — |
| GeForce RTX 5090 | Won't fit | UD-IQ1_M | — |
| Apple M4 Pro | Won't fit | UD-IQ1_M | — |
| Apple M5 | Won't fit | UD-IQ1_M | — |
| CPU only · DDR5-5600 | Won't fit | UD-IQ1_M | — |
Smallest devices that fit it fully
- NVIDIA H200 SXM — Runs, UD-IQ2_XXS, 510–695 tok/s
- NVIDIA B200 — Tight fit, UD-Q3_K_XL, 556–758 tok/s
Run it
Commands for the GeForce RTX 4060 Ti 16GB estimate (UD-IQ1_M).
llama.cpp
llama-server -hf unsloth/Qwen3.5-397B-A17B-GGUF:UD-IQ1_M -c 2048 -ngl 0
Ollama
ollama run qwen3.5
LM Studio
lms get unsloth/Qwen3.5-397B-A17B-GGUF # Then load quant UD-IQ1_M