Kimi-K2 🤝 Anthropic | Blog Post by Justin Wong
Posted by LeveredRecap@reddit | LocalLLaMA | View on Reddit | 9 comments
Congrats to the Moonshot team on Kimi-K2!
Posted by LeveredRecap@reddit | LocalLLaMA | View on Reddit | 9 comments
Congrats to the Moonshot team on Kimi-K2!
Zigtronik@reddit
I would really like you to provide context on why anthropic is mentioned in this post, and why you are congratulating them while linking the blog post? The blog post from a quick readthrough seems like one of the people who worked on it, which that info is hard to find, who is Justin Wong in relation to moonshot?
This post could really benefit from like three more sentences of context.
LeveredRecap@reddit (OP)
Please excuse the rough translation but Justin Wong stated, "most Agent products are nothing without Claude"
More so a win for open-source, above all else
a_beautiful_rhind@reddit
what? How is a proprietary API a "win" for open source?
Zigtronik@reddit
I think the message is something like “The people who made Kimi respect Claude’s agentic abilities, and Kimi does well agentically like it does.”Â
Those are not my claims, just how I am reading it. I think it could be a lot more clear.Â
Zigtronik@reddit
Ah, interesting, I do feel the same as what he said then. Claude code along with Opus and it's proficiencies with tools is far far beyond any other open or closed source agentic products I have used. If that is the benchmark they set themselves against I respect that. I do hope Kimi performs like Opus on those metrics!
datbackup@reddit
Omitting context has apparently become like a religious experience for a lot people… well maybe that’s not exactly it but frankly it doesn’t feel far off. I guess what happens is person B sees person A use little or no context, and get some engagement where the replies “magically” know what person A is talking about, and person B is like “WOOOOW this is so cool, I want to be one of the cool people who talks about something without saying what it is I’m talking about and everyone knows what I’m talking about anyway”
It’s a total plague
And on x for instance it’s a very common engagement farming technique where you can guarantee a bunch of replies saying “what’s the context”
mikael110@reddit
The title makes it sound like there is some partnership or official acknowledgment between Kimi-K2 and Anthropic which is not the case.
The linked blog is just the personal musings of a single developer working on Kimi, he goes out of his way to make it clear he is in no way speaking for the company in any kind of capacity.
The only mention of Anthropic models is in relation to Claude 3.5 sonnet being the model that ushered in the era of LLMs being used and tested increasingly for front-end development. And that Anthropic cutting off Windsurf showcased how reliant AI products are on having access to the top of the line model at the time. And that without a good model the products can't stand on their own.
There is no suggestion that Kimi-K2 and Anthropic is linked in any way.
LeveredRecap@reddit (OP)
I see your point, wasn't my intent to mislead
A Chinese AI open-source startup giving props to a closed-source U.S. AI startup is a pretty uncommon event
The post was more about open-source and closed-source, but I apologize for using the wrong emoji
LeveredRecap@reddit (OP)
Compare that to the DeepSeek moment