Text model · Phi-3.5
Phi-3.5-mini 3.8B requirements
Phi-3.5 family · 3.82B params · released 2024-08-23 · 588.6K Ollama pulls. Minimum to run at Q4_K_M: Apple M1 (8GB).
Quantization sizes
| Quantization | Size on disk |
|---|---|
| Q2_K | 1.6 GB est |
| Q3_K_M | 1.9 GB est |
| Q4_K_M (default) | 2.39 GB |
| Q5_K_M | 2.7 GB est |
| Q6_K | 3.1 GB est |
| Q8_0 | 4.06 GB |
| FP16 | 7.6 GB est |
Lower quant = smaller and faster, slightly lower quality. Q4_K_M is the common default.
Run it
ollama run phi3.5:3.8b llama-cli -hf bartowski/Phi-3.5-mini-instruct-GGUF:Q4_K_M lms get bartowski/Phi-3.5-mini-instruct-GGUF Which devices can run Phi-3.5-mini 3.8B?
Apple Silicon Macs
RAM-only laptops
iPhone & iPad
Android
NVIDIA GPUs
AMD GPUs
FAQ
How much VRAM or RAM does Phi-3.5-mini 3.8B need?
At Q4_K_M, Phi-3.5-mini 3.8B needs about 3.7 GB (weights ~2.39 GB + KV cache + overhead) at a 4k context. At Q8_0 budget ~5.4 GB.
Can Phi-3.5-mini 3.8B run on a laptop?
Yes, Phi-3.5-mini 3.8B fits on a 16 GB machine at Q4_K_M and runs on Apple Silicon or a 12 GB+ GPU comfortably.
Can I use Phi-3.5-mini 3.8B commercially?
Yes. Phi-3.5-mini 3.8B is licensed MIT, which permits commercial use.
Microsoft's 3.8B long-context SLM. 128K context, multilingual (20+ languages). Trained on 3.4T tokens. Strong performance vs similarly-sized models on reasoning and document QA.
Sources
Memory figures are estimates. See methodology.