Phi-4-instruct (reasoning-trained)

Name: Phi-4-instruct (reasoning-trained)
Author: Microsoft Azure

Reasoning modelPhi-4

Release date: 2025-02-27Updated: 2025-02-28Views: 320

Live demoGitHubHugging Face Compare

Parameters

3.8B

Context length

128K

Multilingual

Supported

Reasoning ability

No data

Phi-4-instruct (reasoning-trained) is an AI model published by Microsoft Azure, released on 2025-02-27, for Reasoning model, with 3.8B parameters, and 128K context length, requiring about 7.67GB storage, with a 90.40 score on MATH-500.

Data sourced primarily from official releases (GitHub, Hugging Face, papers), then benchmark leaderboards, then third-party evaluators. Learn about our data methodology

Phi-4-instruct (reasoning-trained)

Model basics

Reasoning traces

Supported

Thinking modes

Thinking modes not supported

Context length

128K tokens

Max output length

No data

Model type

Reasoning model

Modality (in / out)

No data

Release date

2025-02-27

Model file size

7.67GB

MoE architecture

Total params / Active params

3.8B / Not applicable

Knowledge cutoff

No data

Phi-4-instruct (reasoning-trained)

Open source & experience

Code license

Proprietary

Weights license

Proprietary

GitHub repo

N/A

Hugging Face

huggingface.co/microsoft/Phi-4-multimodal-instr…

Live demo

N/A

Phi-4-instruct (reasoning-trained)

Official resources

Paper

DataLearnerAI blog

Phi-4-instruct (reasoning-trained)

API details

API speed

No data

No public API pricing yet.

Phi-4-instruct (reasoning-trained)

Benchmark Results

Phi-4-instruct (reasoning-trained) currently shows benchmark results led by AIME 2024 (46 / 62, score 50), MATH-500 (37 / 45, score 90.40), GPQA Diamond (248 / 274, score 49). This page also consolidates core specs, context limits, and API pricing so you can evaluate the model from benchmark results and deployment constraints together.