Microsoft: Phi 4 Mini Instruct

Phi 4 Mini Instruct is a text-only model from Microsoft with a 131K-token context window and a maximum output of 128K tokens. It does not support tool use, reasoning modes, or structured output, so it operates as a straightforward text-in, text-out completion model. That scope is narrow but defined, and users should not expect it to handle image inputs or integrate into agentic pipelines without additional scaffolding. At $0.08 per million input tokens and $0.35 per million output tokens, it sits at the cheaper end of the market, which is its clearest argument. Benchmark coverage is limited to three benchmarks with a blended score of 4.2, so performance claims should be treated as preliminary rather than well-established. It is most worth shortlisting for cost-sensitive text tasks, such as summarization or classification at volume, where lower capability requirements make the thin benchmark data less of a concern.