Microsoft
Efficient

Phi-3 Small

Compact

7B Phi-3. The middle of the Phi-3 lineup; superseded by Phi-3.5 for new self-hosted work.

Phi-3 Small is a efficient AI model from Microsoft. It costs $0.1500 per million input tokens and $0.6000 per million output tokens (blended $0.2850/M), with a 128,000-token context window.

RatesUSD per million tokens. Blend is 70% input, 30% output, our calculation.

INPUT
$0.1500/M
per million input tokens
OUTPUT
$0.6000/M
per million output tokens
BLENDED 70/30
$0.2850/M
unchanged since 3 May
CONTEXT
128,000
tokens

Hand-maintained from Microsoft's published pricing; no check date on record.

What it is good at
  • Single-GPU
  • Open weights (MIT)
  • 128K context
Typical use cases
  • Legacy Phi-3 small deployments

Benchmarks

vs. best public score
MMLU75.7%
Multitask academic knowledge across 57 subjects. Source: Microsoft (self-reported), MMLU (5-shot), read 2026-11-10.
MATH34.6%
High-school competition math problems. Source: Microsoft (self-reported), MATH (0-shot CoT), read 2026-11-10.
Python function synthesis from docstrings. Source: Microsoft (self-reported), HumanEval (0-shot, phi-3-small-instruct), read 2026-11-10.
Hand-curated from each provider's published reports and public leaderboards. Methodology varies across sources, treat as directional rather than authoritative.

More from Microsoft

See all 8 →

Frequently asked questions

How much does Phi-3 Small cost?

Phi-3 Small costs $0.1500 per million input tokens and $0.6000 per million output tokens, for a blended reference rate of $0.2850 per million tokens.

What is Phi-3 Small's context window?

Phi-3 Small supports up to 128,000 tokens of context in a single request.

What is Phi-3 Small best for?

Phi-3 Small is well suited to Single-GPU, Open weights (MIT) and 128K context.

Who makes Phi-3 Small?

Phi-3 Small is developed and served by Microsoft. It was released in Apr 2024.

Terms used on this page