LongCat
Frontier reasoning models from Meituan
9 followers
Frontier reasoning models from Meituan
9 followers
LongCat is the AI model series from Meituan, featuring powerful, efficient, and open-source models for complex tasks. Its latest release, LongCat-Flash-Thinking, is a 560B parameter MoE reasoning model that sets a new standard for speed and performance.
Flowtica Scribe
Hi everyone!
After acquiring the high-profile AI startup Light Years Beyond in early 2023, Meituan was quiet on the foundational model front for a while. But nearly two years later, the LongCat series is really starting to make some noise.
Following LongCat-Flash earlier this month, they've now open-sourced this powerful new reasoning model, LongCat-Flash-Thinking, and the benchmarks are stunning.
This reasoning model is fast. I mean... surprisingly fast for a 560B parameter model. Its architecture only activates ~27B parameters, and the speed is really impressive.
It's fully open-source with an MIT license. What more do you want? Keep shipping, LongCat! 😼
Fable Fiesta
Damn, that's a really good and fast model...
One thing I don't like is that the API pricing information is hidden under a registration wall, as seems to be the model info (good thing I found it in comment here on product hunt, but if I haven't found you guys here, I would probably have bounced from website). Seriously, there is no reason to keep the pricing and model info behind a login wall.
But otherwise this seems like a solid model, not just solid, it's impressive. I can't wait to test it, and maybe test some fine tunings on it.
Absolutely great job!