Phi 4 Models — Hardware Requirements
7 Phi 4 models from Microsoft and the community, from the smallest that runs in 4.8 GB of VRAM at Q4 up to 14.7B parameters. Every row links to full quantization tables and GPU compatibility.
All Phi 4 Models by Size
| Model | Params | Runs from (Q4) | Context | Publisher | Quant downloads |
|---|---|---|---|---|---|
| Phi 4 Mini Instruct | 3.8B | 4.8 GB | 131K | ||
| Phi 4 Mini Reasoning | 3.8B | 4.8 GB | 131K | ||
| Phi 4 Mini Flash Reasoning | 3.9B | 5.3 GB | 262K | ||
| Phi 4 | 14.7B | 12.5 GB | 16K | ||
| Phi 4 Reasoning Plus | 14.7B | 12.5 GB | 33K | ||
| Phi 4 Reasoning | 14.7B | 12.5 GB | 33K |
Community variants and fine-tunes(1)
| Model | Params | Runs from (Q4) | Context | Publisher | Quant downloads |
|---|---|---|---|---|---|
| Phi 4 Quantized.w8a8 | 14.7B | 12.5 GB | 16K |
How Phi 4 Compares — llmrun Composite Score
Phi 4 is the highest-scoring Phi 4 model with an llmrun Composite Score of 83.9 — #81 among 106 open models. The top proprietary model, Gemini 4 Carbon, scores 158.6. Click a model to see its full benchmark breakdown.
Gemini 4 Carbon · proprietary158.6
GPT 6 Astra · proprietary156.2
Claude Opus 5.5 · proprietary153.0
GPT 6.1 Sol · proprietary151.7
GPT 6.1 Terra · proprietary150.1
Kimi K3.1 Pro142.8
MiMo V2.6 Pro RL133.3
Kimi K3132.7
MiMo V2.6 Flash RL131.9
GLM 5.3130.1
DeepSeek V4.1 Flash129.4
DeepSeek V4 Pro 0813129.2
Phi 483.9
Frequently Asked Questions
- How much VRAM do I need to run a Phi 4 model?
- The smallest Phi 4 model, Phi 4 Mini Instruct, runs from 4.8 GB of VRAM at Q4_K_M (smaller low-bit quantizations start at 1.6 GB with a noticeable quality loss). Larger family members need proportionally more — see the table above for every model.
- Which Phi 4 models can I run on a 16 GB GPU?
- 7 of 7 Phi 4 models fit in 16 GB of VRAM at Q4_K_M, including Phi 4 Mini Instruct, Phi 4, Phi 4 Mini Reasoning.
- What is the most popular Phi 4 model to run locally?
- Phi 4 Mini Instruct is the most downloaded Phi 4 model in local-friendly quantized formats. It runs from 4.8 GB of VRAM at Q4_K_M.
- How do Phi 4 models score on benchmarks?
- Phi 4 leads the family with an llmrun Composite Score of 83.9, ranking #81 among 106 open models, while the top proprietary model, Gemini 4 Carbon, scores 158.6. See the comparison chart above for the full standings.