# Microsoft Phi

16 Hacker News commenters recommended Microsoft Phi and 7 criticized it, November 2025 to November 2026. It was recommended in 11 of 13 months. It meets the Hacker Approved badge rule.

**Verdict:** Small, MIT-licensed Microsoft models good at simple tasks, math, logic and local tool calling; critics say they are weak elsewhere and dated.

## Why people recommend it

- Runs locally in small memory (3 comments): "I've done pretty decent local prose->json extraction using Qwen and Phi and Gemma." ([source](https://news.ycombinator.com/item?id=49480170)) "fits in about 3.5gb of RAM (not vram)" ([source](https://news.ycombinator.com/item?id=48386983)) "I already use Microsoft Phi offline" ([source](https://news.ycombinator.com/item?id=48376068))
- Good for simple tasks (3 comments): "microsofts phi models are very good smaller models under MIT license" ([source](https://news.ycombinator.com/item?id=45842584)) "IMO models like Phi-4 are underrated for many simple tasks" ([source](https://news.ycombinator.com/item?id=46339309)) "the small model punching above its weight via data quality" ([source](https://news.ycombinator.com/item?id=47136127))
- Curated training data gives strong results (2 comments): "Phi was smart (or had a smart training architecture, more precisely), and arguably pushed the conversation forward globally on the value of curated training data." ([source](https://news.ycombinator.com/item?id=49074823)) "the small model punching above its weight via data quality" ([source](https://news.ycombinator.com/item?id=47136127))
- Recent versions better, less sloppy (2 comments): "Gemma and phi have gotten better at it in recent versions" ([source](https://news.ycombinator.com/item?id=46701434)) "better, less sloppy, more natural language style" ([source](https://news.ycombinator.com/item?id=49561773))

## What people warn about

- Feels dated, outclassed by newer models (3 comments): "they are much better than things like the Phi series were" ([source](https://news.ycombinator.com/item?id=46525756)) "Phi was cool for what it was. But it's not 2024 anymore." ([source](https://news.ycombinator.com/item?id=49073599)) "Microsoft Phi 4 hasn't seen an update in a year." ([source](https://news.ycombinator.com/item?id=46269665))
- Loses originality, restricted outputs (2 comments): "That's really good to control if you go for strict limits but it also loses a lot of originality." ([source](https://news.ycombinator.com/item?id=50000749)) "a lobomised model that is probably Phi-4 :(" ([source](https://news.ycombinator.com/item?id=50283049))

## Alternatives hackers recommend

- [Claude Haiku](https://hackerapproved.com/p/claude-haiku.md): 218 recommend
- [OpenAI Luna](https://hackerapproved.com/p/luna.md): 218 recommend
- [Apple Foundation Models](https://hackerapproved.com/p/apple-foundation-models.md): 45 recommend
- [Ministral](https://hackerapproved.com/p/ministral.md): 18 recommend
- [LFM2.5 2.6B](https://hackerapproved.com/p/lfm2-5-2-6b.md): 14 recommend

## Cite as

Hacker Approved, "Microsoft Phi", https://hackerapproved.com/p/phi/ (counts from 1,795,640 Hacker News comments, November 2025 to November 2026).