Phi-4 Mini
Q4 GGUF download: unsloth/Phi-4-mini-instruct-GGUF - 2.49 GB on disk, read 2026-10-17. VRAM figures above are estimates; this is the actual file size.
Intelligence benchmarks
Artificial Analysis indexes - compared with the best open and proprietary models
Intelligence
6.3
AA Index
Coding
3.8
AA Index
Intelligence Index - Phi-4 Mini vs. the field
Best open-weight models (you can run locally) and leading proprietary models for context.
Coding Index comparison
Benchmark data from Artificial Analysis · updated 2026-10-19.
Hosted API cost & speed
What it costs to rent this model instead of running it locally
| Metric | Value |
|---|---|
| Input (per 1M tokens) | $0.08 |
| Output (per 1M tokens) | $0.30 |
| Median speed (hosted API) | 44 tok/s |
Median across hosted providers, measured by Artificial Analysis - updated 2026-10-19. Local llama.cpp speed depends on your hardware — see the hardware pages for estimates.
Standard benchmarks
Performance across standard evaluations
| Benchmark | Score |
|---|---|
| MMLU | 64 |
| HumanEval | 65 |
| MT-Bench | 7.5 |
| GSM8K | 78 |
| MMLU-Pro | 46.5 |
| GPQA Diamond | 33.1 |
| AIME | 6.7 |
| HLE | 4.5 |
| LiveCodeBench | 12.6 |
| IFBench | 21.1 |
| Terminal-Bench | 0 |
Source: Artificial Analysis component evals - updated 2026-10-19. Other scores are carried over from the site's initial dataset - no source on record, not re-verified.
Will it run on your hardware?
Pick your GPU memory - see which quantizations fit, and the cheapest card for the rest
Need an exact figure for your context length? Use the VRAM calculator.
Run it locally
Copy-paste - running in under a minute
ollama run phi-4-mini:3bvllm serve microsoft/Phi-4-mini-instructNew to this? Start with Ollama · serve to many users with vLLM.
Deep dive
Notes, sources, and the full write-up
Phi-4 Mini
Phi-4 Mini brings Microsoft's synthetic-data training approach to a 3.8-billion-parameter model. It targets edge deployment with strong coding performance for its size.
Key features
- Synthetic data training - inherits Phi-4's approach
- 64 MMLU - strong for 3.8B
- 65 HumanEval - best-in-class coding for sub-4B
- MIT license - fully open
How to run
ollama run phi-4-mini:3bFrom the community
What users are saying
“Yes it seems likely that most of these tools are ignoring the 0.75 scaling, thanks for pointing that out @ykim362 ! Will investigate”
“try downgrading trl and transformers I think LossKwargs removed from transformers==4.54.0 https://github.com/sgl-project/sglang/issues/8004issuecomment-3148397838”
“Hi , when i try to deploy the model via a HF Inference Endpoint i get this : [Server Message] Endpoint Failed to Start. With the following details : Exit code: 1.”
Frequently asked
Quick answers to common questions
How much VRAM does Phi-4 Mini need?
Phi-4 Mini with 3B parameters needs approximately 2.5 GB at Q4_K_M quantization. Use our VRAM calculator for an exact estimate.
Is Phi-4 Mini better than other Phi models?
Phi-4 Mini scores 64 on MMLU and 65 on HumanEval. It has 3B parameters with 16,384 context - a strong choice for edge, mobile, chatbots.
What license is Phi-4 Mini under?
Phi-4 Mini is released under the MIT license, making it suitable for most commercial and personal projects.
What hardware runs Phi-4 Mini well?
With 3B parameters, Phi-4 Mini requires adequate VRAM. High-end GPUs like the RTX 4090 (24GB), RTX 5090 (32GB), or Mac Studio with unified memory are good options. Check our hardware directory for specific recommendations.
What is the best quantization for Phi-4 Mini?
Q4_K_M is the recommended sweet spot - ~98% of FP16 quality at ~27% of the size. Q5_K_M (~3 GB) is an option if you have spare VRAM. Use our VRAM calculator to compare.
What models compete with Phi-4 Mini?
Phi-4 Mini competes with other models in its class. Browse our model directory for comparisons, benchmarks, and community reviews to find the best fit.
Compare & pair with
Similar models and recommended hardware
Related models
Recommended hardware
Nearby options
Similar models and compatible hardware by spec