# LongCat-2.0

5 Hacker News commenters recommended LongCat-2.0 and 6 criticized it, November 2025 to November 2026. It was recommended in 3 of 13 months.

**Verdict:** Chinese MoE model that people praise for novel architecture ideas and competitive results with rival models; critics note its weights cannot be downloaded and it is unaudited.

## Why people recommend it

- Novel architecture and N-gram embedding (2 comments): "their contribution far exceeds merely post-training the base model in a different way" ([source](https://news.ycombinator.com/item?id=48727549)) "N-gram embedding seems like an amazing idea" ([source](https://news.ycombinator.com/item?id=48730653))
- Competitive with other top models (2 comments): "Very impressed too - seems roughly on par with Hy3." ([source](https://news.ycombinator.com/item?id=48854353)) "Beats Gwenn and ChatGPT 5.5" ([source](https://news.ycombinator.com/item?id=48791463))

## What people warn about

- Weights not downloadable and unaudited (2 comments): "Nothing can be downloaded from their Huggingface, and given this company's consistent track record, it can basically be considered a scam" ([source](https://news.ycombinator.com/item?id=48729217)) "it's not like any of it is audited" ([source](https://news.ycombinator.com/item?id=48738865))

## Alternatives hackers recommend

- [Claude Opus](https://hackerapproved.com/p/claude-opus.md): 3331 recommend
- [OpenAI GPT](https://hackerapproved.com/p/gpt-5-5.md): 1205 recommend
- [Claude Sonnet](https://hackerapproved.com/p/claude-sonnet.md): 854 recommend
- [DeepSeek V4 Flash](https://hackerapproved.com/p/deepseek-v4-flash.md): 828 recommend
- [Fable](https://hackerapproved.com/p/fable.md): 823 recommend

## Cite as

Hacker Approved, "LongCat-2.0", https://hackerapproved.com/p/longcat-2-0/ (counts from 1,795,640 Hacker News comments, November 2025 to November 2026).