llmboard.ai
Benchmarks
CompareRankings
llmboard.ai
Benchmarks
CompareRankings
HomeModelsPhi 4

Microsoft model product

Phi 4

phi-4 is a state-of-the-art open model built to excel at advanced reasoning, coding, and knowledge tasks. It leverages a blend of synthetic data, filtered web data, academic texts, and supervised fine-tuning for precision, alignment, and safety.

Updated Aug 10, 2026. Default version: Phi 4

Compare
LLMBoard score36.9Phi 4
Coverage80%8 benchmark families
Context window16KTokens
Official input price$0.125Azure API

On this page

  • Specification
  • Capability
  • Benchmarks
  • Arena
  • Pricing
  • Versions
  • About
  • Compare
  • Similar models
  • FAQ

Model specification

Structured fields from the published default version.

Version
Phi 4
Released
Dec 12, 2024
Knowledge cutoff
Jun 1, 2024
Parameters
14.7B
Context window
16K
Max output
16K
Inputs
text
Outputs
text
Open weights
No
License
MIT

Capability profile

This profile uses the latest version under this unique model that has a calculated LLMBoard score. Arena and price are excluded.

Phi 4 category scores

Benchmark results

Published benchmark records for the scored version Phi 4.

13 rows
Columns

Show columns

PhiBench0.6330.0%CAug 7, 2026
Arena Hard0.892668.0%CAug 7, 2026
DROP0.8193037.9%CAug 7, 2026
MATH0.8197174.3%CAug 7, 2026
MGSM0.8193140.0%CAug 7, 2026
LiveBench0.536385.4%CAug 7, 2026
SimpleQA0.044464.4%CAug 7, 2026
HumanEval+0.8457641.3%CAug 7, 2026
MMLU0.84510055.6%CAug 7, 2026
HumanEval0.8467640.0%CAug 7, 2026
IFEval0.663653.1%CAug 7, 2026
MMLU-Pro0.78212936.7%CAug 7, 2026
GPQA0.616423329.7%CAug 7, 2026

Arena results

Preference and agent-evaluation signals from published Arena datasets.

56 rows
Columns

Show columns

textjapanese1981159.2500N/AAug 6, 2026
text style controljapanese2051195.1500N/AAug 6, 2026
textkorean2171151.3319N/AAug 6, 2026
textfrench2211223.4205N/AAug 6, 2026
textspanish2211231.4127N/AAug 6, 2026
textgerman2221222.8523N/AAug 6, 2026
text style controlkorean2231196.2319N/AAug 6, 2026
text style controlspanish2251265.9127N/AAug 6, 2026
text style controlfrench2261265.3205N/AAug 6, 2026
text style controlgerman2281256.3523N/AAug 6, 2026
textindustry mathematical2481253.82,350N/AAug 6, 2026
text style controlindustry mathematical2491278.42,350N/AAug 6, 2026
textmath2531245.82,764N/AAug 6, 2026
textexpert2561202.11,124N/AAug 6, 2026
text style controlexpert2571266.71,124N/AAug 6, 2026
text style controlmath2581264.92,764N/AAug 6, 2026
textindustry legal and government2661241.11,341N/AAug 6, 2026
text style controlindustry legal and government2661298.41,341N/AAug 6, 2026
textchinese2671211.31,503N/AAug 6, 2026
textindustry medicine and healthcare2671201.01,037N/AAug 6, 2026
textcoding2681231.83,305N/AAug 6, 2026
texthard prompts2691219.85,747N/AAug 6, 2026
texthard prompts english2701231.33,804N/AAug 6, 2026
textrussian2711207.92,576N/AAug 6, 2026
text style controlindustry medicine and healthcare2721272.31,037N/AAug 6, 2026
text style controlhard prompts english2731290.23,804N/AAug 6, 2026
textindustry software and it services2741228.85,580N/AAug 6, 2026
text style controlchinese2741267.01,503N/AAug 6, 2026
textindustry business and management and financial operations2751198.62,438N/AAug 6, 2026
textinstruction following2751201.79,162N/AAug 6, 2026
text style controlcoding2751306.43,305N/AAug 6, 2026
textlonger query2771217.52,896N/AAug 6, 2026
text style controlhard prompts2781277.75,747N/AAug 6, 2026
text style controlrussian2781245.72,576N/AAug 6, 2026
textmulti turn2791205.93,517N/AAug 6, 2026
textnon english2791196.810,025N/AAug 6, 2026
text style controlindustry software and it services2791290.15,580N/AAug 6, 2026
textenglish2801231.014,101N/AAug 6, 2026
text style controlindustry business and management and financial operations2801256.22,438N/AAug 6, 2026
text style controllonger query2811266.52,896N/AAug 6, 2026
textoverall2831216.924,126N/AAug 6, 2026
textexclude ties2831100.715,789N/AAug 6, 2026
text style controlinstruction following2831245.39,162N/AAug 6, 2026
text style controlmulti turn2831241.63,517N/AAug 6, 2026
textcreative writing2851182.04,062N/AAug 6, 2026
textindustry life and physical and social science2851206.84,245N/AAug 6, 2026
textindustry writing and literature and language2861197.96,470N/AAug 6, 2026
text style controlenglish2871271.814,101N/AAug 6, 2026
textindustry entertainment and sports and media2881162.54,486N/AAug 6, 2026
text style controlnon english2881233.010,025N/AAug 6, 2026
text style controloverall2891256.224,126N/AAug 6, 2026
text style controlexclude ties2891156.915,789N/AAug 6, 2026
text style controlindustry writing and literature and language2891229.06,470N/AAug 6, 2026
text style controlcreative writing2901210.24,062N/AAug 6, 2026
text style controlindustry life and physical and social science2901264.64,245N/AAug 6, 2026
text style controlindustry entertainment and sports and media2911202.54,486N/AAug 6, 2026

Pricing

Official vendor API PAYG pricing is summarized first. The table then lists individual provider offerings without treating their minimum as the official price.

Official API
$0.125 input, $0.50 output per 1M
Official provider
Azure
Lowest third-party
From $0.07 input, $0.14 output per 1M via OpenRouter
Tracked offerings
3
3 rows
Columns

Show columns

OpenRoutermicrosoft/phi-4global$0.07$0.1416.4KAug 7, 2026
Azure Cognitive Servicesphi-4global$0.125$0.50128KAug 7, 2026
Azurephi-4global$0.125$0.50128KAug 7, 2026

Official prices use only the vendor's configured official Provider and positive standard USD PAYG rates. Third-party offers remain explicitly labeled.

Phi 4 versions

All published versions linked to this unique model. The score columns identify the version used by the current overall ranking.

1 rows
Columns

Show columns

Phi 4Dec 12, 202436.914.7B16K16KNoMIT

What is Phi 4?

A concise description based on the published model registry.

phi-4 is a state-of-the-art open model built to excel at advanced reasoning, coding, and knowledge tasks. It leverages a blend of synthetic data, filtered web data, academic texts, and supervised fine-tuning for precision, alignment, and safety.

Use the benchmark, Arena and pricing sections above as separate evidence. A missing field means the current data snapshot does not support that claim.

Data snapshot: 2026-08-07. Editorial model content is not available in the backend.

Phi 4 vs nearby models

Open a comparison with the three ranked models immediately above and below this model.

Phi 4vsDiffusionGemma 26B A4BPhi 4vsDeepSeek-V3Phi 4vsQwen2 72BPhi 4vsGemini 1.5 FlashPhi 4vsNova MicroPhi 4vsQwen3.5 9B

Models similar to Phi 4

Recommendations prioritize the same model type and family, then the closest published LLMBoard score.

#134-5.5
MI

Phi 4 Mini

Microsoft

31.5 LLMBoard

DetailsCompare
#107+5.5
MI

Phi 3.5 MoE

Microsoft

42.4 LLMBoard

DetailsCompare
#92+10.9
MI

Phi 4 Reasoning

Microsoft

47.9 LLMBoard

DetailsCompare
#144-11.0
MI

Phi 3.5 mini

Microsoft

25.9 LLMBoard

DetailsCompare
#83+13.4
MI

Phi 4 Reasoning Plus

Microsoft

50.3 LLMBoard

DetailsCompare
#44+26.4
MI

MAI Thinking 1

Microsoft

63.4 LLMBoard

DetailsCompare

FAQ

Common questions about Phi 4.

When was Phi 4 released?

Phi 4's default version was released on Dec 12, 2024.

How much does Phi 4 cost?

Phi 4's official API price is $0.125 per million input tokens and $0.50 per million output tokens via Azure. The lowest tracked third-party offer starts at $0.07 input and $0.14 output via OpenRouter.

Who created Phi 4?

Phi 4 is published under Microsoft in the model registry.

What is the context window for Phi 4?

The default version has a 16K token context window.

Is Phi 4 open weight?

No. The default version is not marked as having publicly available weights.

How many API providers offer Phi 4?

3 published provider offerings are linked to the default version.

What models should I compare Phi 4 with?

Nearby ranked alternatives include DiffusionGemma 26B A4B, DeepSeek-V3, Qwen2 72B.

Rankings

OverallCodingText ArenaPricing

Modalities

Image GenerationVideo GenerationSpeech-to-TextEmbeddings

Benchmarks

All BenchmarksReasoningMathCoding

Vendors

All VendorsOpenAIAnthropicGoogle
llmboard.aiCopyright 2026 llmboard.ai