unsloth
GLM 4 9B 0414
GLM 4 9B 0414 is a 9.40B dense glm4 model from unsloth. Configured context length is 32,768 tokens. License: mit. Quant sizes are the GGUF file bytes published on Hugging Face when a file exists.
Parameters9.4B
Context32K
Attentiongqa
Licensemit
40 layers · 2 KV heads · head dim 128 · GGUF repo · config
Quant file sizes
| Quant | File size | Source |
|---|---|---|
| BF16 | 17.5 GB | Hugging Face file bytes |
| UD-Q8_K_XL | 11.2 GB | Hugging Face file bytes |
| Q8_0 | 9.31 GB | Hugging Face file bytes |
| UD-Q6_K_XL | 8.28 GB | Hugging Face file bytes |
| Q6_K | 7.70 GB | Hugging Face file bytes |
| UD-Q5_K_XL | 6.54 GB | Hugging Face file bytes |
| Q5_K_M | 6.57 GB | Hugging Face file bytes |
| Q5_K_S | 6.24 GB | Hugging Face file bytes |
| UD-Q4_K_XL | 5.78 GB | Hugging Face file bytes |
| Q4_K_M | 5.74 GB | Hugging Face file bytes |
| Q4_K_S | 5.36 GB | Hugging Face file bytes |
| IQ4_NL | 5.09 GB | Hugging Face file bytes |
| IQ4_XS | 4.92 GB | Hugging Face file bytes |
| Q4_1 | 5.60 GB | Hugging Face file bytes |
| Q4_0 | 5.10 GB | Hugging Face file bytes |
| UD-Q3_K_XL | 4.79 GB | Hugging Face file bytes |
| Q3_K_M | 4.63 GB | Hugging Face file bytes |
| Q3_K_S | 4.28 GB | Hugging Face file bytes |
| UD-IQ3_XXS | 3.94 GB | Hugging Face file bytes |
| UD-Q2_K_XL | 3.92 GB | Hugging Face file bytes |
| Q2_K | 7.60 GB | Hugging Face file bytes |
| UD-IQ2_XXS | 3.27 GB | Hugging Face file bytes |
| UD-IQ1_M | 3.09 GB | Hugging Face file bytes |
| UD-IQ1_S | 2.98 GB | Hugging Face file bytes |
On common machines
| Device | Verdict | Quant | Speed |
|---|---|---|---|
| GeForce RTX 4060 Ti 8GB | Tight fit | Q5_K_S | 23–31 |
| GeForce RTX 3060 12GB | Tight fit | Q8_0 | 19–26 |
| GeForce RTX 4060 Ti 16GB | Runs | UD-Q8_K_XL | 13–17 |
| GeForce RTX 4090 | Runs | BF16 | 29–39 |
| GeForce RTX 5090 | Runs | BF16 | 51–70 |
| Apple M4 Pro | Runs | UD-Q8_K_XL | 12–16 |
| Apple M5 | Tight fit | Q8_0 | 8.1–11 |
| CPU only · DDR5-5600 | Runs on CPU | BF16 | 2.6–3.5 |
Smallest devices that fit it fully
- GeForce RTX 3050 6GB — Tight fit, Q3_K_S, 19–26 tok/s
- GeForce RTX 2060 6GB — Tight fit, Q3_K_S, 38–51 tok/s
- GeForce GTX 1660 Super — Tight fit, Q3_K_S, 38–51 tok/s
- GeForce GTX 1660 Ti — Tight fit, Q3_K_S, 32–44 tok/s
- Intel Arc A380 — Tight fit, Q3_K_S, 21–28 tok/s
- GeForce RTX 5060 Ti 8GB — Tight fit, Q5_K_S, 35–48 tok/s
Run it
Commands for the GeForce RTX 4060 Ti 16GB estimate (UD-Q8_K_XL).
llama.cpp
llama-server -hf unsloth/GLM-4-9B-0414-GGUF:UD-Q8_K_XL -c 8192 -ngl 40
No matching name in the Ollama library snapshot.
LM Studio
lms get unsloth/GLM-4-9B-0414-GGUF # Then load quant UD-Q8_K_XL