wait... so you're telling me five randoms just put $300k into training an AI model just to see what happens?? and they ended up beating OpenAI and Anthropic for a fraction of the cost no PhDs and with under $4M raised, they built a model that does the work of bankers, consultants and lawyers: -they taught it 88 kinds of the real tasks those people do every day - now you type ONE prompt and it builds the deck in PowerPoint, the financial model in Excel and the memo in Word - in two weeks they're releasing

wait... so you're telling me five randoms just put $300k into training an AI model just to see what happens?? and they ended up beating OpenAI and Anthropic for a fraction of the cost no PhDs and with under $4M raised, they built a model that does the work of bankers, consultants and lawyers: -they taught it 88 kinds of the real tasks those people do every day - now you type ONE prompt and it builds the deck in PowerPoint, the financial model in Excel and the memo in Word - in two weeks they're releasing the model, so any company can run it on its own servers and keep its data at home it's called Ultramarine 1 Flash... try it in the Feather app and lemme know what you get

We are releasing Ultramarine 1 Flash, a frontier-class model built for knowledge work. It is open-source and achieves similar performance to Fable 5, while being 60x cheaper. It was post-trained by a team of 5 for ~$300k all-in. Available on API now, with weights releasing in 2

[Video 3](blob:https://x.com/e4d199fd-6814-475e-adfe-27487802c1c2)

01:17

Paid partnership