Microsoft: Phi 4

Microsoft: Phi 4

microsoft · Released Jan 10, 2025 Professional
Intelligence #8 / 571
82.0 Our Score
Speed #244 / 266
38.1 tokens / sec
Input #170 / 571
$0.065 per 1M tokens
Output #155 / 571
$0.140 per 1M tokens
Context #424 / 571
16,384 tokens

Analysis Summary

Microsoft Phi 4 is a small-footprint model priced at $0.065 input and $0.14 output per million tokens, with a strong math index of 18 and a coding index of 11.2 relative to its intelligence index of 10.4. This suggests it punches above its weight on structured reasoning tasks, though its agentic index is very low and it lacks tool use or vision support.

For businesses, the 16K context window is a significant constraint, ruling it out for long-document analysis or extended agentic sessions. It is best suited to focused, short-context tasks such as math problem solving, code generation, or structured data processing where context length is not a bottleneck.

At this price point, it offers good value for narrow, well-scoped tasks. Teams looking for a cheap coding or math assistant for short-context work will find it capable, but those needing broader capability or longer context should look elsewhere.

Assessed June 6, 2026

Editorial notes

Microsoft Phi 4 is a compact, cheap model with strong math and coding scores relative to its size, but a very small 16K context window limits its use in document-heavy business workflows.

Rankings consider pricing, capabilities, benchmarks, and real-world applicability and are refreshed as new models launch. Feedback?

Performance Profile

Intelligence2.4Technical1.3Value7.5Content2.6
Intelligence 2.4/10
Technical 1.3/10
Content 2.6/10
Value 7.5/10

How Microsoft: Phi 4 compares

Microsoft: Phi 4 ranks #304 of 374 AI models we track for overall intelligence, #218 of 311 for coding, #278 of 286 for agentic tasks. Its 16K-token context window is larger than 26% of the models we list. At $0.07 per million input tokens it is cheaper than 70% of comparable models.

About Microsoft: Phi 4

Microsoft Research Phi-4 is designed to perform well in complex reasoning tasks and can operate efficiently in situations with limited memory or where quick responses are needed. At 14 billion..

Performance Indices

Source: Artificial Analysis

10.4 Intelligence Index
11.2 Coding Index
3.8 Agentic Index
18 Math Index

Benchmark Scores

Intelligence

GPQA Diamond 57.5% Graduate-level scientific reasoning
HLE 4.1% Humanity's Last Exam
MMLU Pro 71.4% Multi-task language understanding
MATH 500 81% Mathematical problem-solving
AIME 14.3% Competition mathematics
AIME 2025 18% Competition mathematics (2025)
SciCode 26% Scientific computing

Technical

LiveCodeBench 23.1% Live coding evaluation
TerminalBench Hard 3.8% Agentic terminal tasks

Content

IFBench 23.5% Instruction following

Benchmark data from Artificial Analysis and Hugging Face

How does Microsoft: Phi 4 stack up?

Compare side-by-side with other professional models.

Compare Models

Model Information

OpenRouter ID microsoft/phi-4
Providermicrosoft
Release Date January 10, 2025
Context Length16,384 tokens
Max Completion16,384 tokens
Status Active

Pricing

Token Type Cost per 1M tokens Cost per 1K tokens
Input $0.07 $0.000065
Output $0.14 $0.000140

Live Performance

Live endpoint metrics, refreshed every 30 minutes.

100%
Avg Uptime
254ms
Best Latency (TTFT)
51 tok/s
Best Throughput
2/2
Active Endpoints
Available via: NextBit, DeepInfra

Leaderboard Categories

Frequently asked questions about Microsoft: Phi 4

How much does Microsoft: Phi 4 cost?

Microsoft: Phi 4 costs $0.07 per million input tokens and $0.14 per million output tokens.

What is the context window of Microsoft: Phi 4?

Microsoft: Phi 4 has a context window of 16,384 tokens (16K).

Is Microsoft: Phi 4 good for coding?

On our coding benchmark index, Microsoft: Phi 4 ranks #218 of 311 models, placing it in the broader range of the field for code generation and debugging.

Who created Microsoft: Phi 4?

Microsoft: Phi 4 is developed by Microsoft and was released on January 10, 2025.