Phi-4 Multimodal Instruct

Microsoft·Pending·Released 2025-02-26textvisionaudio

Context128k

Max output4.096k

Input / 1M-

Output / 1M-

Cached in-

Benchmarks19

Benchmarks

Cited public sources · rank is exact-benchmark, apples-to-apples

AI2D

82.3% ↗

BLINK

61.3% ↗

ChartQA

#1/281.4% ↗

DocVQA

#1/293.2% ↗

InfoVQA

#1/272.7% ↗

InterGPS

48.6% ↗

MathVista testmini

62.4% ↗

MMBench (dev-en)

86.7% ↗

MMMU

#4/555.1% ↗

MMMU-Pro (standard/vision)

38.5% ↗

OCRBench

84.4% ↗

POPE

85.6% ↗

ScienceQA Visual

#1/297.5% ↗

Speech AI2D

68.9% ↗

Speech ChartQA

69% ↗

Speech DocVQA

87.3% ↗

Speech InfoVQA

63.7% ↗

TextVQA (val)

75.6% ↗

Video-MME (16 frames)

55% ↗

Compare

vs GPT-6 Astra vs GPT-5.6 Sol vs GPT-5.6 Terra vs GPT-5.6 Luna vs Claude Fable 5.1 vs Claude Opus 5 vs Claude Sonnet 5 vs Claude Haiku 4.5

More from Microsoft

MAI-Thinking-1 Phi-4 Phi-4 Reasoning Phi-4 Mini Reasoning

Access Phi-4 Multimodal Instruct and every other model through one endpoint with automatic failover: Respan gateway.