Phi-4-multimodal-instruct
Multimodal
Zero-eval
#1ScienceQA Visual
#1BLINK
#1InterGPS
+2 more
by Microsoft
About
Phi-4-multimodal-instruct is a multimodal language model developed by Microsoft. It achieves strong performance with an average score of 72.0% across 15 benchmarks. It excels particularly in ScienceQA Visual (97.5%), DocVQA (93.2%), MMBench (86.7%). The model shows particular specialization in general tasks with an average performance of 75.8%. It supports a 256K token context window for handling large documents. The model is available through 1 API provider. As a multimodal model, it can process and understand text, images, and other input formats seamlessly. It's licensed for commercial use, making it suitable for enterprise applications. Released in 2025, it represents Microsoft's latest advancement in AI technology.
Pricing Range
Input (per 1M)$0.05 -$0.05
Output (per 1M)$0.10 -$0.10
Providers1
Timeline
AnnouncedFeb 1, 2025
ReleasedFeb 1, 2025
Knowledge CutoffJun 1, 2024
Specifications
Training Tokens5.0T
Capabilities
Multimodal
License & Family
License
MIT
Benchmark Performance Overview
Performance metrics and category breakdown
Overall Performance
15 benchmarks
Average Score
72.0%
Best Score
97.5%
High Performers (80%+)
7Performance Metrics
Max Context Window
256.0KAvg Throughput
25.0 tok/sAvg Latency
1msTop Categories
general
75.8%
vision
69.7%
math
62.4%
Benchmark Performance
Top benchmark scores with normalized values (0-100%)
Ranking Across Benchmarks
Position relative to other models on each benchmark
ScienceQA Visual
Rank #1 of 1
#1Phi-4-multimodal-instruct
97.5%
DocVQA
Rank #14 of 26
#11Grok-2 mini
93.2%
#12Pixtral Large
93.3%
#13DeepSeek VL2
93.3%
#14Phi-4-multimodal-instruct
93.2%
#15GPT-4o
92.8%
#16Nova Lite
92.4%
#17DeepSeek VL2 Small
92.3%
MMBench
Rank #2 of 7
#1Qwen2.5 VL 72B Instruct
88.0%
#2Phi-4-multimodal-instruct
86.7%
#3Qwen2.5 VL 7B Instruct
84.3%
#4Phi-3.5-vision-instruct
81.9%
#5DeepSeek VL2 Small
80.3%
POPE
Rank #2 of 2
#1Phi-3.5-vision-instruct
86.1%
#2Phi-4-multimodal-instruct
85.6%
OCRBench
Rank #4 of 7
#1Qwen2.5 VL 7B Instruct
86.4%
#2Qwen2-VL-72B-Instruct
87.7%
#3Qwen2.5 VL 72B Instruct
88.5%
#4Phi-4-multimodal-instruct
84.4%
#5DeepSeek VL2 Small
83.4%
#6DeepSeek VL2
81.1%
#7DeepSeek VL2 Tiny
80.9%
All Benchmark Results for Phi-4-multimodal-instruct
Complete list of benchmark scores with detailed information
ScienceQA Visual ScienceQA Visual benchmark | vision | multimodal | 0.97 | 97.5% | Self-reported |
DocVQA DocVQA benchmark | vision | multimodal | 0.93 | 93.2% | Self-reported |
MMBench MMBench benchmark | general | text | 0.87 | 86.7% | Self-reported |
POPE POPE benchmark | general | text | 0.86 | 85.6% | Self-reported |
OCRBench OCRBench benchmark | general | text | 0.84 | 84.4% | Self-reported |
AI2D AI2D benchmark | general | text | 0.82 | 82.3% | Self-reported |
ChartQA ChartQA benchmark | general | multimodal | 0.81 | 81.4% | Self-reported |
TextVQA TextVQA benchmark | vision | multimodal | 0.76 | 75.6% | Self-reported |
InfoVQA InfoVQA benchmark | vision | multimodal | 0.73 | 72.7% | Self-reported |
MathVista MathVista benchmark | math | text | 0.62 | 62.4% | Self-reported |
Showing 1 to 10 of 15 benchmarks