This was my thought too. But somewhat revealing of the mindset involved in an awful lot of the AI "industry".
First, start with a pile of money and the vague notion of an AI tool. Usefulness will come later, surely, definitely, before we run out of pile of money.
Is it the AI industry or just VC in general? YC quite famously invests in founders, not ideas. Other accelerators are similar. They start with a pile of money and the vague notion of a startup idea. Usefulness comes later, after however many pivots, before they run out of the pile of money.
Quick summary: Apply for up to $6K reimbursement per month for up to 6 months for Llama usage (wherever you use it). Incorporated startups with less than $10 million USD in funding are eligible. (disclosure: I work at Meta as Llama Partner Engineer)
Do I count as the founder? I use contractors from Eastern Europe to do frontend, does that count?
Overall seems like a very niche offering, considering $6K is peanuts these days. You can get more by applying for any of Microsoft, Google, or Amazon startup programs. MSFT for instance straight up gives > $100K in cloud credits when you are funded (and if not, how do you pay your developer?)
> Join us for the opportunity to receive cloud reimbursements of up to $6,000 USD per month for up to six months, technical resources, and a vibrant community
Thanks :) A/VR is not my thing. Hopefully they will do something in the future. Tertiary Education is in crisis thanks to AI, but it’s also going to supercharge it and that’s where research is needed.
If you come up with a great idea, then Facebook can either steal it from you or buy you out without them having to spend the funds to test all of the potential options.
Facebook hasn't had a good idea for over a decade (and even that one was trash), so they need a little help now and again.
Meta doesn't have a business providing straight up API usage. This is a new market, if they can get startups relying on Llama hosting, then there's no reason this can't be a whole new market.
Yes. You will just need to document how much of your OpenRouter bill was for Llama usage (like provide screenshots if its not broken down on your invoices or receipts).
Yes. That's what the "does it have to be legally incorporated" question is implicitly contrasting with, by my reading. (Not against some putative "non-legally incorporated company".)
You don’t have to be incorporated for Azure startup program. Sole proprietor LLC works fine. It’s kind of an insane deal… $125k credits for a $150 LLC and a YouTube demo.
I want to know the same thing, but the other day I heard an interesting stat that ChatGPT is the #4 top site by traffic, just behind Instagram and ahead of X. We mustn’t let all of that attention go elsewhere shall we?
I love open weight (and better open source) LLMs and wish Llama all the best! But God help Meta if they have to pay startups to entice them to use their open weights model.
I'm surprised they're leveraging a png for it instead of webp or some other compression format. Mainly surprised that compression of blog assets isn't automated for them.
I am not affiliated with the program, but I suspect an LLC would count as incorporated, and therefore an LLC with a solo dev (that has raised less than $10m) would qualify.
I've only used Scout, but I found it more than acceptable as a "GPT-4 or better" level model for coding questions. I've not stress tested it in any way because I didn't need to. So it's fine! It's just not special.
> There is a 700 million monthly active users limit.
...so? Who cares? Why does it matter if big megacorporations cannot use it for free? I sure as hell don't care.
If you want to complain about the license a better target is the annoying advertising clause:
> i. If you distribute or make available the Llama Materials (or any derivative works thereof), or a product or service (including another AI model) that contains any of them, you shall (A) provide a copy of this Agreement with any such Llama Materials; and (B) prominently display “Built with Llama” on a related website, user interface, blogpost, about page, or product documentation. If you use the Llama Materials or any outputs or results of the Llama Materials to create, train, fine tune, or otherwise improve an AI model, which is distributed or made available, you shall also include “Llama” at the beginning of any such AI model name.
Heh, I had a go at doing something in this space with AI a while back: finding properties via a prompt. It didn't end up working very well. Way too much hallucinations.
Nice, what was it called is it still up? I'd love to hear why it didnt work out.. As of now, my product isnt using prompts and analyzes deals using a stacking ensemble model I made on recently trained data
love it. any llm can be made to perform reliably and accurately which is the biggest pre-requisite when it comes to creating an "AI Agent". I think this gives people the opportunity to start somewhere because they can leverage multi-pass prompting frameworks like TSCE to scale: https://github.com/AutomationOptimization/tsce_demo. despite the fact that "llama isn't the best"
Exactly what leads to inaccurate output in LLM's. The semantic interpretation of each individual token isn't the same between us and it. "Interpretation", we likely define accuracy and reliability pretty closely the same you and I, but that interpretation is where we differ. As for my "definition" it's in the repo. I'm not funded by anyone, don't get paid, and have no product to sell. So if you're genuinely interested in discussing, I'm all for it!
> Applications for the initial cohort close on May 30, 2025 at 6:00 pm PT.