So Apple is about to pay OpenAI 1 B usd pr year for what moonshot is giving for free?
Kimi K2 Thinking, a SOTA open-source trillion-parameter reasoning model
41–50 of 442 posts
Re: Kimi K2 Thinking, a SOTA open-source trillion-parameter reasoning model
#42Four independent Chinese companies released extremely good open source models in the past few months (DeepSeek, Qwen/Alibaba, Kimi/Moonshot, GLM/Z.ai). No American or European companies are doing that, including titans like Meta. What gives?
Re: Kimi K2 Thinking, a SOTA open-source trillion-parameter reasoning model
#43Would be nice if this were on AWS bedrock or google vertex for data residency reasons.
The non-thinking Kimi K2 is on Vertex AI, so it's just a matter of time before it appears there. Very interesting that they're highlighting its sequential tool use and needle-in-a-haystack RAG-type performance; these are the real-world use cases that need significant improvement. Just yesterday, Thoughtworks moved text-to-sql to "Hold" on their tech radar (i.e. they recommend you stop doing it).
Thanks, I didn't realize Thoughtworks was staying so up-to-date w/ this stuff.
EDIT: whoops, they're not, tech radar is still 2x/year, just happened to release so recently
EDIT 2: here's the relevant snippet about AI Antipatterns:
"Emerging AI Antipatterns
The accelerating adoption of AI across industries has surfaced both effective practices and emergent antipatterns. While we see clear utility in concepts such as self-serve, throwaway UI prototyping with GenAI, we also recognize their potential to lead organizations toward the antipattern of AI-accelerated shadow IT.
Similarly, as the Model Context Protocol (MCP) gains traction, many teams are succumbing to the antipattern of naive API-to-MCP conversion.
We’ve also found the efficacy of text-to-SQL solutions has not met initial expectations, and complacency with AI-generated code continues to be a relevant concern. Even within emerging practices such as spec-driven development, we’ve noted the risk of reverting to traditional software-engineering antipatterns — most notably, a bias toward heavy up-front specification and big-bang releases. Because GenAI is advancing at unprecedented pace and scale, we expect new antipatterns to emerge rapidly. Teams should stay vigilant for patterns that appear effective at first but degrade over time and slow feedback, undermine adaptability or obscure accountability."
Re: Kimi K2 Thinking, a SOTA open-source trillion-parameter reasoning model
#44It's good to see more competition, and open source, but I'd be much more excited to see what level of coding and reasoning performance can be wrung out of a much smaller LLM + agent as opposed to a trillion parameter one. The ideal case would be something that can be run locally, or at least on a modest/inexpensive cluster. The original mission OpenAI had, since abandoned, was to have AI benefit all of humanity, and…
"open source" means there should be a script that downloads all the training materials and then spins up a pipeline that trains end to end. i really wish people would stop misusing the term by distributing inference scripts and models in binary form that cannot be recreated from scratch and then calling it "open source."
"open source" has come to mean "open weight" in model land. It is what it is. Words are used for communication, you are the one misusing the words.
You can update the weights of the model, continue to train, whatever. Nobody is stopping you.
Re: Kimi K2 Thinking, a SOTA open-source trillion-parameter reasoning model
#45It's good to see more competition, and open source, but I'd be much more excited to see what level of coding and reasoning performance can be wrung out of a much smaller LLM + agent as opposed to a trillion parameter one. The ideal case would be something that can be run locally, or at least on a modest/inexpensive cluster. The original mission OpenAI had, since abandoned, was to have AI benefit all of humanity, and…
"open source" means there should be a script that downloads all the training materials and then spins up a pipeline that trains end to end. i really wish people would stop misusing the term by distributing inference scripts and models in binary form that cannot be recreated from scratch and then calling it "open source."
> binary form that cannot be recreated from scratch
Back in my day, we called it "freeware"
Re: Kimi K2 Thinking, a SOTA open-source trillion-parameter reasoning model
#46Four independent Chinese companies released extremely good open source models in the past few months (DeepSeek, Qwen/Alibaba, Kimi/Moonshot, GLM/Z.ai). No American or European companies are doing that, including titans like Meta. What gives?
The Chinese are doing it because they don't have access to enough of the latest GPUs to run their own models. Americans aren't doing this because they need to recoup the cost of their massive GPU investments.
Re: Kimi K2 Thinking, a SOTA open-source trillion-parameter reasoning model
#47It's good to see more competition, and open source, but I'd be much more excited to see what level of coding and reasoning performance can be wrung out of a much smaller LLM + agent as opposed to a trillion parameter one. The ideal case would be something that can be run locally, or at least on a modest/inexpensive cluster. The original mission OpenAI had, since abandoned, was to have AI benefit all of humanity, and…
"open source" means there should be a script that downloads all the training materials and then spins up a pipeline that trains end to end. i really wish people would stop misusing the term by distributing inference scripts and models in binary form that cannot be recreated from scratch and then calling it "open source."
I'd agree but we're beyond hopelessly idealistic. That sort of approach only helps your competition who will use it to build a closed product and doesn't give anything of worth to people who want to actually use the model because they have no means to train it. Hell most people can barely scrape up enough hardware to even run inference.
Reproducing models is also not very ecological in when it comes down to it, do we really all need to redo the training that takes absurd amounts of power just to prove that it works? At least change the dataset to try and get a better result and provide another datapoint, but most people don't have the knowhow for it anyway.
Nvidia does try this approach sometimes funnily enough, they provide cool results with no model in hopes of getting people to buy their rented compute and their latest training platform as a service...
Re: Kimi K2 Thinking, a SOTA open-source trillion-parameter reasoning model
#48Earlier quoted context omitted.
"open source" means there should be a script that downloads all the training materials and then spins up a pipeline that trains end to end. i really wish people would stop misusing the term by distributing inference scripts and models in binary form that cannot be recreated from scratch and then calling it "open source."
> binary form that cannot be recreated from scratch Back in my day, we called it "freeware"
You have more rights over a freely licensed binary file than over a freeware file.
Re: Kimi K2 Thinking, a SOTA open-source trillion-parameter reasoning model
#49It's good to see more competition, and open source, but I'd be much more excited to see what level of coding and reasoning performance can be wrung out of a much smaller LLM + agent as opposed to a trillion parameter one. The ideal case would be something that can be run locally, or at least on a modest/inexpensive cluster. The original mission OpenAI had, since abandoned, was to have AI benefit all of humanity, and…
> I'd be much more excited to see what level of coding and reasoning performance can be wrung out of a much smaller LLM + agent
Well, I think you are seeing that already? It's not like these models don't exist and they did not try to make them good, it's just that the results are not super great.
And why would they be? Why would the good models (that are barely okay at coding) be big, if it was currently possible to build good models, that are small?
Of course, new ideas will be found and this dynamic may drastically change in the future, but there is no reason to assume that people who work on small models find great optimizations that frontier models makers, who are very interested in efficient models, have not considered already.
Re: Kimi K2 Thinking, a SOTA open-source trillion-parameter reasoning model
#50Four independent Chinese companies released extremely good open source models in the past few months (DeepSeek, Qwen/Alibaba, Kimi/Moonshot, GLM/Z.ai). No American or European companies are doing that, including titans like Meta. What gives?
Meta gave up on open weight path after DeepSeek.