HT Sherpa

A model is the software that produces AI answers. This guide compares models and tools. New to this? Choose a setup path first. Use the lists below when you want more detail.

New words, in plain language
Local / cloud
Local runs on your computer. Cloud runs on computers managed online.
Model / tokens
A model produces AI answers. Tokens are small pieces of text; some services charge for the tokens used.
Account / identity
An account is a sign-in. Your cloud identity is the account you use for online services.
Connector / permissions
A connector links an AI tool to another service. Permissions are the access you allow.

Field Guide · Week of 5 October 2026

Phi-4

Microsoft · United States · Microsoft

Small, sharp, MIT-licensed. Strong at exams and short reasoning. Less warm as a letter-writer.

Where it runs
Runs on your computer
Memory
12–16 GB (4-bit)
Size
14B dense
License
MIT
Hugging Face downloads
427k
Hugging Face likes
2.3k
Card last updated
14 July 2026
On your computer · #8United States · #6

Why it is on this week’s list

The cleanest US license on a model that still thinks. Good second model next to Gemma.

Use it when

Math, code snippets, tight answers on 16 GB.

Skip it when

Long, human letters — Gemma or Qwen will sound more like you.

Published scores

Scores as of 17 August 2026. A blank test is not a zero. We just will not invent it.

GPQA Diamond
56.1*

Microsoft Phi-4 card · 2025-01

Older card. Still the honest public number.

SWE-Bench Verified
—

No public number we will stand behind.

LiveCodeBench
—

No public number we will stand behind.

MMLU / MMLU-Pro
84.8*

Microsoft Phi-4 card · 2025-01

AIME
—

No public number we will stand behind.

Humanity’s Last Exam
—

No public number we will stand behind.

What these numbers actually mean
Walk me through installing thisHugging Face cardAll lists