unsloth
gemma 4 31B it
gemma 4 31B it is a 30.7B dense gemma4 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.
Parameters30.7B
Context256K
Attentionhybrid
Licenseapache-2.0
60 layers · 16 KV heads · head dim 256 · 50 sliding layers · GGUF repo · config
Quant file sizes
| Quant | File size | Source |
|---|---|---|
| BF16 | 57.2 GB | Hugging Face file bytes |
| UD-Q8_K_XL | 32.6 GB | Hugging Face file bytes |
| Q8_0 | 30.4 GB | Hugging Face file bytes |
| UD-Q6_K_XL | 25.6 GB | Hugging Face file bytes |
| Q6_K | 23.5 GB | Hugging Face file bytes |
| UD-Q5_K_XL | 20.4 GB | Hugging Face file bytes |
| Q5_K_M | 20.2 GB | Hugging Face file bytes |
| Q5_K_S | 19.7 GB | Hugging Face file bytes |
| UD-Q4_K_XL | 17.5 GB | Hugging Face file bytes |
| Q4_K_M | 17.1 GB | Hugging Face file bytes |
| Q4_K_S | 16.2 GB | Hugging Face file bytes |
| IQ4_NL | 16.1 GB | Hugging Face file bytes |
| IQ4_XS | 15.2 GB | Hugging Face file bytes |
| Q4_1 | 17.8 GB | Hugging Face file bytes |
| Q4_0 | 16.1 GB | Hugging Face file bytes |
| UD-Q3_K_XL | 14.3 GB | Hugging Face file bytes |
| Q3_K_M | 13.7 GB | Hugging Face file bytes |
| Q3_K_S | 12.3 GB | Hugging Face file bytes |
| UD-IQ3_XXS | 11.0 GB | Hugging Face file bytes |
| UD-Q2_K_XL | 11.0 GB | Hugging Face file bytes |
| UD-IQ2_XXS | 7.95 GB | Hugging Face file bytes |
On common machines
| Device | Verdict | Quant | Speed |
|---|---|---|---|
| GeForce RTX 4060 Ti 8GB | Partial offload | Q4_K_M | 3.3–4.5 |
| GeForce RTX 3060 12GB | Tight fit | UD-IQ2_XXS | 12–16 |
| GeForce RTX 4060 Ti 16GB | Tight fit | Q3_K_S | 9.2–13 |
| GeForce RTX 4090 | Tight fit | Q5_K_S | 22–30 |
| GeForce RTX 5090 | Tight fit | UD-Q6_K_XL | 28–38 |
| Apple M4 Pro | Tight fit | IQ4_XS | 7.4–10 |
| Apple M5 | Tight fit | UD-IQ2_XXS | 5.1–6.9 |
| CPU only · DDR5-5600 | Runs on CPU | UD-Q5_K_XL | 1.6–2.2 |
Smallest devices that fit it fully
- GeForce RTX 2080 Ti — Tight fit, UD-IQ2_XXS, 27–37 tok/s
- GeForce GTX 1080 Ti — Tight fit, UD-IQ2_XXS, 21–29 tok/s
- GeForce RTX 5070 — Tight fit, UD-IQ2_XXS, 22–30 tok/s
- GeForce RTX 4070 Ti — Tight fit, UD-IQ2_XXS, 17–23 tok/s
- GeForce RTX 4070 Super — Tight fit, UD-IQ2_XXS, 17–23 tok/s
- GeForce RTX 4070 — Tight fit, UD-IQ2_XXS, 17–23 tok/s
Run it
Commands for the GeForce RTX 4060 Ti 16GB estimate (Q3_K_S).
llama.cpp
llama-server -hf unsloth/gemma-4-31B-it-GGUF:Q3_K_S -c 4096 -ngl 60
Ollama
ollama run gemma
LM Studio
lms get unsloth/gemma-4-31B-it-GGUF # Then load quant Q3_K_S