Skip to main content

Phi-4 Multimodal Instruct

MicrosoftPhiOpen WeightMIT · Commercial OK

Description

Phi-4-multimodal-instruct is a lightweight (5.57B parameters) open multimodal foundation model that leverages research and datasets from Phi-3.5 and 4.0. It processes text, image, and audio inputs to generate text outputs, supporting a 128K token context length. Enhanced via SFT, DPO, and RLHF for instruction following and safety.

Release Date
2025-02-26
Parameters
5.6B
Context Length
Modalities
image, text

Capability Radar

18
general
13
coding
32
reasoning
19
scienceest.
26
agents
85
multimodal

Science uses a reasoning proxy when dedicated science benchmarks are unavailable.

Rankings

Domain#RankScoreSource
Code Ranking416
14.0
AA
General Ranking440
21.0
AA
Math Reasoning228
39.0
AA
Multimodal Ranking28
82.0
LS
Science453
17.0
AA

Benchmark Scores (LLM Stats)

3d

BLINK61.3%SR

General

MMMU55.1%SR
MMMU-Pro38.5%SR

Image To Text

DocVQA93.2%SR
OCRBench84.4%SR
TextVQA75.6%SR

Math

MathVista62.4%SR
InterGPS48.6%SR

Multimodal

ScienceQA Visual97.5%SR
MMBench86.7%SR
POPE85.6%SR
AI2D82.3%SR
ChartQA81.4%SR
InfoVQA72.7%SR
Video-MME55.0%SR

AA Evaluation Indices

Intelligence Index
4.5
Math 500
0.7
Mmlu Pro
0.5
Gpqa
0.3
Livecodebench
0.1
Scicode
0.1
Aime
0.1
Hle
0.0

LLM Stats Category Scores

Image To Text
80
Multimodal
70
Reasoning
70
Vision
70
Math
60
Spatial Reasoning
60
Healthcare
60
3d
60
General
50

Pricing

Input PriceFree
Output PriceFree
Blended Price (3:1)Free

Speed

Tokens/sec16.0
Time to First Token1.34s
Time to Answer1.34s

Provider Price Ranking

Provider Price Ranking

4 providers

Cheapest: NanoGPTMost Expensive: evroc
ProviderInputOutput
1NanoGPTCheapest
$0.07
$0.11
2Azure Cognitive Services
$0.08
$0.32
3Azure
$0.08
$0.32
4evroc
$0.24
$0.47

Compare pricing across different API providers for this model.

External Sources