Phi 4 by Microsoft scores 19.3 out of 100 on the Neura Intelligence Index, ranking #304 among tracked AI models. This composite score is computed from 2 individual benchmarks across coding, math, reasoning, and language categories.

Phi 4 costs $0.07/1M input tokens and $0.14/1M output tokens via API. At a typical 3:1 input-to-output ratio with 1M tokens/day usage, the estimated monthly cost is approximately $3.

Yes, Phi 4 has open weights available under the mit license. You can download and self-host the model.

Phi 4 supports a context window of 16K tokens, allowing it to process moderate-length documents and conversations in a single request.

Phi 4 achieves its highest score on GPQA Diamond with a score of 56.1%. This benchmark evaluates 448 graduate-level science questions (biology, chemistry, physics) designed to be hard for non-experts.