unsloth
Olmo 3 7B Instruct
Olmo 3 7B Instruct is a 7.30B dense olmo2 model from unsloth. Configured context length is 65,536 tokens. License: apache-2.0. Quant sizes are the GGUF file bytes published on Hugging Face when a file exists.
Parameters7.3B
Context64K
Attentionhybrid
Licenseapache-2.0
32 layers · 32 KV heads · head dim 128 · 24 sliding layers · GGUF repo · config
Quant file sizes
| Quant | File size | Source |
|---|---|---|
| BF16 | 13.6 GB | Hugging Face file bytes |
| UD-Q8_K_XL | 8.68 GB | Hugging Face file bytes |
| Q8_0 | 7.23 GB | Hugging Face file bytes |
| UD-Q6_K_XL | 6.33 GB | Hugging Face file bytes |
| Q6_K | 5.58 GB | Hugging Face file bytes |
| UD-Q5_K_XL | 4.86 GB | Hugging Face file bytes |
| Q5_K_M | 4.85 GB | Hugging Face file bytes |
| Q5_K_S | 4.73 GB | Hugging Face file bytes |
| UD-Q4_K_XL | 4.26 GB | Hugging Face file bytes |
| Q4_K_M | 4.16 GB | Hugging Face file bytes |
| Q4_K_S | 3.96 GB | Hugging Face file bytes |
| IQ4_NL | 3.93 GB | Hugging Face file bytes |
| IQ4_XS | 3.74 GB | Hugging Face file bytes |
| Q4_1 | 4.33 GB | Hugging Face file bytes |
| Q4_0 | 3.94 GB | Hugging Face file bytes |
| UD-Q3_K_XL | 3.58 GB | Hugging Face file bytes |
| Q3_K_M | 3.40 GB | Hugging Face file bytes |
| Q3_K_S | 3.08 GB | Hugging Face file bytes |
| UD-IQ3_XXS | 2.74 GB | Hugging Face file bytes |
| UD-Q2_K_XL | 2.82 GB | Hugging Face file bytes |
| Q2_K | 5.41 GB | Hugging Face file bytes |
| UD-IQ2_XXS | 2.10 GB | Hugging Face file bytes |
| UD-IQ1_M | 1.90 GB | Hugging Face file bytes |
| UD-IQ1_S | 1.81 GB | Hugging Face file bytes |
On common machines
| Device | Verdict | Quant | Speed |
|---|---|---|---|
| GeForce RTX 4060 Ti 8GB | Tight fit | UD-Q4_K_XL | 24–32 |
| GeForce RTX 3060 12GB | Tight fit | Q8_0 | 16–22 |
| GeForce RTX 4060 Ti 16GB | Runs | UD-Q8_K_XL | 12–16 |
| GeForce RTX 4090 | Runs | BF16 | 29–40 |
| GeForce RTX 5090 | Runs great | BF16 | 52–71 |
| Apple M4 Pro | Tight fit | BF16 | 7.9–11 |
| Apple M5 | Tight fit | Q8_0 | 7.0–9.5 |
| CPU only · DDR5-5600 | Runs on CPU | BF16 | 2.6–3.6 |
Smallest devices that fit it fully
- GeForce RTX 3050 6GB — Tight fit, UD-IQ3_XXS, 18–25 tok/s
- GeForce RTX 2060 6GB — Tight fit, UD-IQ3_XXS, 36–49 tok/s
- GeForce GTX 1660 Super — Tight fit, UD-IQ3_XXS, 36–49 tok/s
- GeForce GTX 1660 Ti — Tight fit, UD-IQ3_XXS, 31–42 tok/s
- Intel Arc A380 — Tight fit, UD-IQ3_XXS, 20–27 tok/s
- GeForce RTX 5060 Ti 8GB — Tight fit, UD-Q4_K_XL, 37–50 tok/s
Run it
Commands for the GeForce RTX 4060 Ti 16GB estimate (UD-Q8_K_XL).
llama.cpp
llama-server -hf unsloth/Olmo-3-7B-Instruct-GGUF:UD-Q8_K_XL -c 8192 -ngl 32
No matching name in the Ollama library snapshot.
LM Studio
lms get unsloth/Olmo-3-7B-Instruct-GGUF # Then load quant UD-Q8_K_XL