Frontier models like ChatGPT, Gemini, and Claude are very powerful, but that capability comes at a steep cost. What if you could get the same level of artificial intelligence and reasoning from an AI model, but without shelling out millions in tokens? That’s the goal of Together AI.
As trillions of dollars flow into the AI boom, it has led some to question whether the cost structure makes sense. A McKinsey report published earlier this year found global AI spending is on pace to hit $7 trillion by 2030, largely at the hands of giant AI firms and hyperscalers.
As these companies look to get a return on their investments, the costs of accessing frontier AI models clearly has to increase. That’s one reason why more than three-quarters of organizations are expected to shift to using open source AI, according to McKinsey.
Together AI, which was founded four years ago at the beginning of the generative AI boom by Vipul Ved Prakash, Percy Liang, and Ce Zhang, finds itself at the intersection of GenAI’s increasing capabilities and its increasing costs.
“What began as a conviction about the future of AI has grown faster than even our most optimistic projections,” wrote Prakash, the CEO, in a recent blog post. “We believed this generational technology should be open and abundantly available, not controlled by a handful of companies.”
The San Francisco-based company has developed a platform that allows customers to access and run open weight models like DeepSeek, Nemotron, MiniMax, Kimi, and GLM. Together AI doesn’t develop AI models; rather it provides the hardware and software–and all the software glue, like kernels, compilers, and optimizers–that is necessary to get use out of them.
Companies that use the open-weight models can save a substantial amount of money, while also giving them freedom to customize and fine-tune the models to meet their specific needs. According to Together AI, moving away from a proprietary model to an open-weight one can lower costs by 6x to 60x, without hurting performance. Together AI also allows customers to use frontier models along with open-weight models.
Vipul Ved Prakash, CEO and co-founder of Together AI
The business model has found traction with paying customers. Companies like Cognition, Decagon, Eleven Labs, Cursor, and Suno are paying Together AI at an annual run rate of $1.15 billion to get access to its platform.
Last week, the company announced that it raised $800 million in a Series C round led by Aramco Ventures with participation from Vista Equity Partners, General Catalyst, Emergence Capital, NVIDIA, March Capital, Pegatron, and S Ventures. The round valued the company at $8.3 billion.
In addition to the equity capital, the company has secured commitments for over 500 MW of compute capacity thanks to its investors. The compute capacity will be used to support the company’s increased infrastructure use, which it expects to grow 50x in the next five years.
The company and its investors are making a big bet that the future of AI will be open. Prakash is confident that it’s the right one.
“History shows that the biggest technology shifts are won by open ecosystems that make innovation cheaper, faster and available to everyone,” he said in a press release. “We believe AI will follow the same path.”
The post Frontier AI Model Power But 60X Cheaper? It’s Possible, Says Together AI appeared first on AIwire.
