Models Model details

Microsoft

Microsoft: Phi 4

microsoft/phi-4

[Microsoft Research](/microsoft) 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 parameters, it was trained on a mix of high-quality synthetic datasets, data from curated websites, and academic materials. It has undergone careful improvement to follow instructions accurately and maintain strong safety standards. It works best with English language inputs. For more information, please see Phi-4 Technical Report

Input
$0.07

/ 1M tokens

Output
$0.14

/ 1M tokens

Context window
16.38K

tokens

Limits

Context

16,384

tokens

Max output

14,745

90% of the context window

Parameters

14,659,507,200

Model facts

Author

Microsoft

Tokenizer

Other

Modality

Text -> Text

License

MIT

Knowledge cutoff

2024-06-30

Created

Jan 10, 2025

Hugging Face

microsoft/phi-4

Publication status

Hugging Face: verifiedLicense: publishedParameter count: published

Modalities

Input · TextOutput · Text

Generation controls

Frequency penaltyLogit biasMax tokensMin pPresence penaltyRepetition penaltyResponse formatSeedStopStructured outputsTemperatureTop kTop p

Capability scores

5.9
Intelligence index

Better than 4% of models compared