Open Source Generative AI Startup Together Raises $20M

Generative AI startup Together has raised $20 million in a seed funding round led by Lux Capital. Together is building the foundation of support for an open source ecosystem of enterprise generative AI models and tools and suggesting it could be the Linux for the industry.

Generating Together

Together describes its plan as a “decentralized cloud” for AI. The startup essentially wants to gather and organize the generative AI models and services that would interest businesses and help them embed the technology in their products. For now, that means working on a cloud platform to fine-tune and run these models. Together claims its platform will be far less expensive than major cloud services like Google Cloud, Azure, and AWS while still being more secure and easier to customize. That’s a view shared in a leaked Google memo about how tech giants will struggle against open-source generative AI projects, further explained in our Synthedia newsletter.

“In founding Together, we were driven by the belief that open and decentralized alternatives to closed systems were going to be important — and possibly critical for business and society,” Together CEO Vipul Ved Prakash explained in a blog post about the funding. “Our aim is to help create open models that outrival closed models and establish open-source as the default way to incorporate AI. Leveraging research in distributed optimization, we have built a specialized cloud platform for large models that efficiently scales training and inference. In the coming months we plan to open up access to this platform, enabling rapid customization and coupling of foundation models with production tasks. Open models will give developers and organizations greater ability to understand, inspect, and utilize AI, without vendor lock-in and with strong privacy protections.”

Together recently started bringing together open-source AI models as part of its RedPajama project, starting with a 1.2 trillion token dataset that enterprises can use for pre-training models available for licensing. As the mascot above suggests, RedPajama successfully reproduces Meta’s LLaMA model. The name and outfit reference the popular children’s book Llama Llama Red Pajama. Stability AI’s Vicuna model and Stanford’s (quickly ended) Alpaca beat RedPajama to the LLaMA-based model development, but this is one of the first commercially available examples. Together plans to come out with many more RedPajama-based LLMs and AI tools.

Follow @voicebotaiFollow @erichschwartz

Meta Introduces Large Language Model LLaMA as a Competitor for OpenAI

Stanford Closes Meta LLaMA-Based Alpaca Generative AI Demo Over Safety and Cost Problems

Perplexity AI Raises $25.6M and Launches Conversational Search Engine iOS App