Phi-4 is Microsoft's language model with a 16K context window and up to 16K output tokens, available from 3 providers, starting at $0.07 / 1M input and $0.14 / 1M output. A small language model from Microsoft trained on high-quality synthetic and curated datasets, excelling at reasoning and STEM tasks despite its compact size.
Specifications
Canonical IDmicrosoft-phi-4
TypeLanguage
StatusActive
CreatorMicrosoftMicrosoft
Providers
Context Window16K tokens
Max Output16K tokens
Input ModalitiesText
Output ModalitiesText
Parameters14.7B
HuggingFace Likes2,248
HuggingFace Downloads (30d)874,552
HuggingFace Downloads (all-time)11,669,853
Release Date · 2 years ago
Knowledge Cutoff · 2 years ago
Benchmarks
Intelligence Index
#483
Math Index
#226
MMLU-Pro
#214
GPQA
#362
HLE
#503
LiveCodeBench
#258
AIME
#115
IFBench
#399
Time to First Token
#507
SciCode
#385
MATH-500
#108
AIME 2025
#226
LCR
#478
TerminalBench Hard
#304
TAU2
#412
Output TPS
#219

Capabilities

Input1/5
Text✓
Image·
Audio·
Video·
PDF·
Output1/5
Text✓
Image·
Audio·
Video·
Embedding·
Capabilities3/13
Reasoning·
Adaptive Reasoning·
Function Calling✓
Parallel Function Calling·
Structured Outputs✓
Native JSON Schema✓
Web Search·
URL Context·
Computer Use·
Code Execution·
File Search·
Prompt Caching·
Assistant Prefill·

Pricing by Provider

US Dollar ($)
Per 1M tokens
ProviderStandard
Input
$ / 1M
Output
$ / 1M
Azure AI Foundry logo
Azure AI Foundry
microsoft:phi4
$0.125$0.5
DeepInfra logo
DeepInfra
deepinfra/microsoft/phi-4
$0.07$0.14
OpenRouter logo
OpenRouter
microsoft/phi-4
$0.07$0.14

Cost Calculator

US Dollar ($)
Preset:

Cheapest Instances to Run It

Cloud GPU instances that can host Phi-4, ranked by cheapest on-demand price. The model needs about 35 GB of GPU memory at FP16 precision (estimated from its parameter count), so treat the fit as guidance rather than a guarantee.

All clouds
FP16 (full precision)
US Dollar ($)
Instance
Cloud
GPU
VRAM
Price
Cheapest region
Standard_NP10sAzureAMD Alveo U250 FPGA (64GB)64 GB$1.65/hr
g6e.xlargeAWSL40S45 GB$1.86/hr
Standard_NV32as_v4AzureAMD Radeon Instinct MI2564 GB$1.86/hr
7 more instances can run Phi-4
Unlock the full ranked list and FP8 / INT4 quantization with a CloudPrice subscription.

Versions

VersionReleasedContextInput / 1MOutput / 1MStatus
Phi-4 Mini Instruct131K$0.075$0.300Deprecated
Phi-416K$0.070$0.140Current
Phi-4 Mini————Available
Phi-4 Multimodal————Available
Phi-4 Eagle————Available
Phi-4 Mini MM————Available
Phi-4 Mini Reasoning—131K——Available
Phi-4 Multimodal Instruct—131K$0.080$0.320Available
Phi-4 Reasoning—33K——Available
Phi-4 Reasoning Plus————Available
Phi-3 Mini————Available

Model IDs

azure_ai/Phi-4
deepinfra/microsoft/phi-4
microsoft-phi-4
microsoft/phi-4
openrouter/microsoft/phi-4
phi-4
phi-4-mini
phi-4-multimodal
publishers/microsoft/models/phi4