MicrosoftReleased on Feb 1, 2025

Phi 4 Mini: Benchmarks, Pricing & Size

Phi 4 Mini is a language model from Microsoft, released in February 2025.

Phi 4 Mini Instruct is a lightweight (3.8B parameters) open model built upon synthetic data and filtered web data, focusing on high-quality reasoning. It supports a 128K token context length and is enhanced for instruction adherence and

Phi 4 Mini model size

Phi 4 Mini has 3.8 billion parameters and was trained on 5 trillion tokens. See how it compares to other models in the same parameter range.

ParametersTraining tokens
3.8B
5Ttokens
1302× tokens-to-params ratio
Small (3–10B)
3.8B
1B7B70B405B

Phi 4 Mini API

API access coming soon

Phi 4 Mini will be available through our gateway shortly.

Phi 4 Mini examples

Recent arena outputs from Phi 4 Mini, picked from the highest-ranked matchups.

Phi 4 Mini license

Phi 4 Mini is released under the MIT license, which permits commercial use, has 3.8B parameters, has a knowledge cutoff of June 2024.

License
MIT
Commercial use allowed
Parameters
3.8B
Knowledge cutoff
June 2024

MIT License - allows commercial use

FAQ

Common questions about Phi 4 Mini.

When was Phi 4 Mini released?

Phi 4 Mini was released on February 1, 2025 by Microsoft. This is the official Phi 4 Mini release date tracked on LLM Stats.

How big is Phi 4 Mini?

Phi 4 Mini has 3.8 billion parameters. It was trained on 5.0 trillion tokens. It ships as an open-weight model, so you can download and run it on your own hardware.

Who created Phi 4 Mini?

Phi 4 Mini was created by Microsoft.

What is the license for Phi 4 Mini?

Phi 4 Mini is released under the MIT license. This is an open-source / open-weight license that permits self-hosting.

What is the knowledge cutoff date for Phi 4 Mini?

Phi 4 Mini has a knowledge cutoff of June 2024, meaning it was trained on data up to that point and may not know about events after it.

Where is the Phi 4 Mini paper or technical report?

Phi 4 Mini has a paper or technical report available at https://arxiv.org/pdf/2503.01743. Use that source for architecture, training, release and evaluation details.