Phi-4 mini (3.8B)

This model is not part of the weekly automatic sync. The figures below were confirmed on 2026-08-13.

3.8B high-speed inference compact model

Phi-4 mini (3.8B) — Microsoft. Context window 128,000 tokens. API pricing (per 1M tokens): Input $0.06, Output $0.18. License: Open-Weight.

Benchmarks · API pricing (per 1M tokens)

ProviderMicrosoft
TierMicro
Context window128,000 tokens
Max output4,096 tokens
LicenseOpen-Weight
API pricing (per 1M tokens) · Input$0.06
API pricing (per 1M tokens) · Output$0.18
Compare in the full catalog