Re:The path to ubiquitous AI (17k tokens/sec)
by dakolli on Friday February 20, 2026 @06:22AM
(#47086576)
try here, I hate llms but this is crazy fast. https://chatjimmy.ai/
[ Parent
| on HN ]
Re:The path to ubiquitous AI (17k tokens/sec)
by machiaweliczny on Friday February 20, 2026 @08:12AM
(#47087596)
We need that for this chinese 3B model that think 45s for hello world but also solves math.
Re:The path to ubiquitous AI (17k tokens/sec)
by Bolwin on Saturday February 21, 2026 @11:29AM
(#47102234)
Nanbeige. Yeah this seems ideal for models that scale test time compute
Re:The path to ubiquitous AI (17k tokens/sec)
by bmacho on Friday February 20, 2026 @06:30AM
(#47086663)
"447 / 6144 tokens"
"Generated in 0.026s • 15,718 tok/s"
This is crazy fast. I always predicted this speed in ~2 years in the future, but it's here, now.
Re:The path to ubiquitous AI (17k tokens/sec)
by Lalabadie on Friday February 20, 2026 @06:31AM
(#47086678)
The full answer pops in milliseconds, it's impressive and feels like a completely different technology just by foregoing the need to stream the output.
Re:The path to ubiquitous AI (17k tokens/sec)
by FergusArgyll on Friday February 20, 2026 @06:50AM
(#47086842)
Because most models today generate slowish, they give the impression of someone typing on the other end. This is just <enter> -> wall of text. Wild
Re:The path to ubiquitous AI (17k tokens/sec)
by Serenacula on Saturday February 21, 2026 @04:51AM
(#47099146)
Do we know anything about the method?