---
format: "aidr-story-markdown/v1"
id: "a6f0cc316480792c8feedd4f849db0f8e7fa874d46f04381cae157482b4eab2c"
canonical_url: "https://aidr.today/a6f0cc31?lang=en"
title: "Janas: Run LLMs larger than RAM by streaming experts from SSD"
lang: "en"
requested_lang: "en"
available_langs: ["en","vi"]
translation_fallback: null
fallback_fields: []
published_at: "2026-10-12T22:05:43.000Z"
category: "Open Source"
topics: ["llm","open-source"]
source_urls: ["https://github.com/prabanta-dev/janas","https://news.ycombinator.com/item?id=50063717"]
summary: "Janas runs large LLMs by streaming experts from an SSD to exceed RAM limits, achieving 37 tokens per second from Qwen3-Next-80B on a 32 GB laptop."
---

# Janas: Run LLMs larger than RAM by streaming experts from SSD

> [Open the canonical story](<https://aidr.today/a6f0cc31?lang=en>)

**Published:** 2026-10-12T22:05:43.000Z
**Category:** Open Source
**Topics:** llm, open\-source

## Summary

Janas runs large LLMs by streaming experts from an SSD to exceed RAM limits, achieving 37 tokens per second from Qwen3\-Next\-80B on a 32 GB laptop\.

## Sources

- [Story source](<https://github.com/prabanta-dev/janas>)
- [Discussion](<https://news.ycombinator.com/item?id=50063717>)