Subquadratic Raises $29M for Long-Context AI Architecture

A new AI company, Subquadratic, just raised $29 million in seed funding. They've also launched SubQ 1M-Preview, an AI model that could change how large language models work. Here's the thing: traditional AI models struggle with long information because their computing needs skyrocket. Subquadratic claims their new architecture fixes this. Their model scales linearly, not quadratically, making it much more efficient. What's interesting is SubQ 1M-Preview can handle up to 12 million tokens of context. This significantly reduces computing power needed by nearly 1,000 times compared to other advanced AI. They're also rolling out new products, including an API for developers and tools for coding and research. The bottom line: This technology aims to make AI more powerful and affordable for complex tasks, potentially unlocking new applications for everyone.