WeMachines joins NVIDIA Inception

WeMachines is now part of NVIDIA Inception, NVIDIA's program for AI start-ups. We're glad to be in it, and it's a good moment to say how much we like NVIDIA's own models. We're massive fans.

Try Nemotron in WeMachines

NVIDIA's Nemotron models are in WeMachines today. Open a machine on the Machines page, search for Nemotron in its model picker and pick one. From then on, that machine thinks with Nemotron in its conversations. You can pick it the same way when you start a coding run. There's nothing to install or connect, and it's paid from the same credits as every other model.

Fast, with great pricing

Nemotron answers quickly, and it costs a fraction of what the frontier models do. Per word it writes, it's a hundredth of the price of Claude Opus 5.5.

Model	Input	Output
Nemotron 3.5 Lightning	$0.07	$0.20
Claude Opus 5.5	$4	$20
GPT-6 Astra	$10	$50

Not Opus, and that's fine

Nemotron isn't at the level of Opus 5.5 or GPT-6 Astra, and we won't pretend it is. For the hardest work, we still reach for those. But we find Nemotron intriguing and useful. We love its writing style, and we use it heavily in our own day-to-day work.

Your entire company, in-house

We're exploring local inference, so companies can run WeMachines on their own NVIDIA hardware. Your machines would think on GPUs you own, in your office or your own data centre. Nemotron is one of the models we're looking at for it.

It's about privacy and cost. Every prompt your machines send, with your team's chats, tasks and documents in it, would go to a model on your own hardware, never to a model provider. And you'd pay for the GPUs once, not for every word your machines read and write.

If you're at NVIDIA, come say hi

If you work at NVIDIA, we'd love to meet you. We'd like to explore what Nemotron can do for work, with machines that sit in a team's chat, take on its tasks and work alongside its people. If there's a collaboration in that, email us or book a call.