Jailbroken PS5s Run Local AI Models briefing

Jailbroken PS5 Consoles Now Run Local AI Models at Speeds Over 100 Tokens Per Second

A new fork of the PS5LM project enables jailbroken PlayStation 5s to run local AI models natively, leveraging GPU and expanded RAM to achieve over 100 tokens per second on smaller models.

Edited by The Intel Brief Desk · 16 Oct 2026, 17:01 · 1 min read

Key takeaways

A fork of the PS5LM project enables jailbroken PlayStation 5 consoles to run local AI language models natively.
This version uses the PS5’s GPU and accesses more RAM, allowing it to handle larger models and achieve faster processing speeds.
8-27B model runs at about 20 tokens per second on the PS5.

What we know

A fork of the PS5LM project enables jailbroken PlayStation 5 consoles to run local AI language models natively. This version uses the PS5’s GPU and accesses more RAM, allowing it to handle larger models and achieve faster processing speeds. 8-27B model runs at about 20 tokens per second on the PS5.

Why it matters

This development expands the capabilities of the PlayStation 5 beyond gaming, turning it into a platform for running large language models locally. By leveraging the console’s GPU and increased memory access, the forked PS5LM can run more complex AI models faster than before. This could open new possibilities for AI applications on consumer hardware without relying on cloud services.

What is still unknown

No independent tests have confirmed the accuracy or performance claims. Details about which languages the models support and when this fork might be more widely available remain unclear.

Get the daily brief by email.

Jailbroken PS5 Consoles Now Run Local AI Models at Speeds Over 100 Tokens Per Second

What we know

Why it matters

What is still unknown

Sources

Related