141–150 of 238 posts
Earlier quoted context omitted.
Opus 4.5 became significantly cheaper directly per token
You are right I forgot about that ! I think my point still stands - price per token is not decreasing for frontier capabilities, in fact it's increasing.
Earlier quoted context omitted.
> I am optimistic Kimi K open models soon will outperform Opus models Hard to outperform the model you distill...
Is that true? If the distillation is not lossy and the model runs much faster due to less resource consumption, then it may outperform.
Earlier quoted context omitted.
Words on the street is that xAI will buy cursor.
Yeah for 10-60 BILLION. which again makes this even stupider. For this amount of money you can rebuild cursor and everything else on the market, and with the rest of 9-59 Billion, you just hire experts in coding and let them code real high quality code examples. And then you just use your existing grok pipeline and just add this functionality. This xAI stuff has to be run by idiots
And if you combine a shitton of data with a lot of compute, large userbase and good engineers, you have a pretty good chance of doing something interesting.
Any reason why they indexed on Kimi K2.5 model? I have tried many open-source ones in Opencode, and, in my experience (standard backend development, Java, Python, Spring, etc) Qwen3.6 is SO MUCH BETTER that's shocking. Kimi can't even get most tool calling arguments right.
There's a lead time on models, and there's some tuning gotchas they probably already figured out with Kimi, so they weren't ready to just drop everything and switch. I'm sure they will switch models eventually.
Together with SpaceXAI, we're training a significantly larger model from scratch, using 10x more total compute.
With Colossus 2's million H100-equivalents and our combined data and training techniques, we expect this to be a major leap in model capability.Any reason why they indexed on Kimi K2.5 model? I have tried many open-source ones in Opencode, and, in my experience (standard backend development, Java, Python, Spring, etc) Qwen3.6 is SO MUCH BETTER that's shocking. Kimi can't even get most tool calling arguments right.
Earlier quoted context omitted.
>& now they're still losing all of their users to Claude Code and Codex. Why pay for Cursor when I can use GLM 5.1, Kimi K2.6, MiniMax M2.7, Xiaomi MiMo V2.5 Pro and Deepseek v4 for cheap and use whatever harness I want, including Claude Code. It's not like Cursor harness is the best out there. And even if I want to edit the code, I don't need to run the agent harness in an IDE.
these are in the trillion parameters range, not sure it's actually that cheap to have at a reasonable speed without quality degradation & without like.. your own DGX B200
Earlier quoted context omitted.
Yeah for 10-60 BILLION. which again makes this even stupider. For this amount of money you can rebuild cursor and everything else on the market, and with the rest of 9-59 Billion, you just hire experts in coding and let them code real high quality code examples. And then you just use your existing grok pipeline and just add this functionality. This xAI stuff has to be run by idiots
Buy "Cursor", not "Cursor's IP". This means brand, users, and a shitton of data. And if you combine a shitton of data with a lot of compute, large userbase and good engineers, you have a pretty good chance of doing something interesting.
You could literaly just give your compute away for free for a year to pull people in.
Make an API Endpoint for free with the caviat that they are allowed to use the data for traing, what everyone else does too.
Earlier quoted context omitted.
Isn't a large user base and the data collected from those users a moat of sorts?
A moat is when you have something other's can't easily get. Every MAG 7 / FAANG company already has more users and more data... That's not a moat. That's traction.
> Composer 2.5 is built on the same open-source checkpoint as Composer 2, Moonshot's Kimi K2.5. Really nice to see they're giving credit to the company and I am optimistic Kimi K open models soon will outperform Opus models
> I am optimistic Kimi K open models soon will outperform Opus models Hard to outperform the model you distill...
Distillation helps with world knowledge and things like that.