Honestly, Spot feels like magic from the very beginning - and only get's better the more you use it.
I genuinely can't wait to see that first spark of awe from our customers as they go from first chat > to first assistant > to daily use.
I still remember the first time I got everything to work end-to-end - and the fact I am about to share that same exact experience with our customers makes me feel so grateful.
Over the next few weeks, I'll personally be delivering, and setting up Spot devices for each of our Beta customers, and I couldn't be more excited!
Thank you! And yes - I am working right now to get the cost as low as I can - the core components cost us around $2500 (chipset + carrier board) - the rest of the costs are still TBD as we are working with suppliers for V1 manufacturing. But we set a price of $4.5k as the estimate, is to give us plenty of room just in case our manufacturing costs go up.
So - hopefully we will be able to get these devices to customers for much less than $4.5k - but I'll have a solid price in the next month or so!
And here's the specs on the device itself:
https://t.co/jsY2q8cfCL
Yes - believe me - if I could sell it for half as much I would!
Right now were still in beta-testing phase - and we are working with manufacturers and suppliers to get a full build-out together.
So right now we are saying "$4.5k" - just to provide us head-room in case of, for example, the cost of RAM goes up again.
Also - as we scale up manufacturing and start producing in greater bulk, our costs will be able to come down and we will be able to lower the price.
Again, I wan't to get these devices and all the software we have built around them into the hands of every consumer and business that wants them. But the last thing I'd ever want to do is advertise a low price to get pre-orders, and then have to raise it later.
@DSweetTexas@BrainFrame_AI But my favorite that I am using right now is the "nemotron" models from NVIDIA - perfect for tool-calling and multi-modal and really fast. Perfect for most of our Beta-testers use-cases.
Thanks Daniel! Founder here - right now while were in the beta-testing phase we are testing at least 5 different models at any one time - here's a screenshot of what's currently loaded on my device.
As new models come out were constantly updating what's on the device - however the plan for V1 of Spot is three fold:
1. Bring your own model(s) - this allows advanced users to customize which model is being ran on the device.
2. Regularly updated preset of opensource models (ie Gemma, Qwen, Nemo, etc.)
3. A custom tailored "Brain-Frame Model" that is built specifically for the Spot device (think of this like Cursor's 'Composer' model)
Solomon, CEO here - That sounds really cool. I genuinely believe that local edge AI devices are the solution for so many of the problems in the industry right now. The reason we're packaging everything together, the device, the software, the apps, is to make it usable by the everyday person on a device that is specifically built for running AI.
So you can kind of think of it like Claude in a box!
Its a customized AI computing device. Under the hood is: 64GB of unified memory, 1TB of storage, delivers 275+ TOPs, and uses less power than ___.
We built a set of customized tools, harnesses, and operating system that is tailored specifically delivering Claude intelligence and UX completely using quantized opensource models that run completely on the device.
Are you worried about the Utah Data Center?
Or concerned about the water it takes to run big models?
If so โ thereโs a better way. Try Local AI โ intelligence that runs right on your device, zero cloud dependency, zero water waste.
Ready to rethink how AI works? Letโs talk.
https://t.co/YrszeeCzwe
Youโre completely right, apologies for not having the hardware specs on the website or anywhere yet. Iโll make sure to get them posted - weโre just working with manufacturers right now to pin down exactly what the V1 product will look like when itโs shipped later this year.
But as of right now, here are the core specs for our prototypes:
- RAM: 64GB
- Storage: 1TB
- TOPs: 250+
Let me know if youโre looking for more specific stats like latency, TTFT, or TPS.
I don't know if this is a skill issue or something... But I've been playing with OpenClaw a ton for the past 4ish months. I feel like lately I'm spending more time troubleshooting issues with it and telling it what it's doing wrong than I am actually getting valuable use from it.