Technological giants want to boast about AI models of trillion parameters that require mass and expensive GPU clusters. But Speedboat It takes a different approach.
Palo Alto -based starting starts that it has invented a new type of AI architectural model that is deliberately small and specific. The models are so small that they are trained with a low -fee GPU game of less than $ 100,000 in total, Fastino says.
The method attracts attention. Fastino has secured $ 17.5 million in funding of seeds led by Khosla Ventures, the first investor of Openai’s Venture, tells TechCrunch exclusively.
This brings the total funding of the start to about $ 25 million. It set $ 7 million last November in a pre-a-dive round led by Microsoft’s VC Arm M12 and Insight partners.
“Our models are faster, more accurate and cost a fraction to train while overcoming flagship models in specific tasks,” says Ash Lewis, CEO and co -founder of Fastino.
Fastino has created a suite of small models that sells business customers. Each model focuses on a specific job that a company may need, such as the sensitive data it emits or to summarize the corporate documents.
Fastino does not yet reveal early measurements or users, but it says that its performance is wowing early users. For example, because they are so small, its models can offer a whole answer to a single symbolic, Lewis told TechCrunch, showing the technology that gives a detailed answer at the same time in milliseconds.
TechCrunch event
Berkeley, ca
|
June 5
Book now
It’s still a little early to say if Fastino’s approach will catch. The AI Enterprise is full, with companies such as Cohere and Databricks, which also formulate the AI that excelled in certain tasks. And SATA model manufacturers that focus on the business, including the anthropogenic and Mistral, also offer small models. It is also no secret that the future of AI genetics for the business is likely in smaller, more focused linguistic models.
Time can say, but an early vote of confidence by Khosla certainly does not hurt. At the moment, Fastino says he is focusing on building a peak AI team. He targeted the researchers at the top AI Labs that are not obsessed with building the larger model or hitting the benchmarks.
“Our recruitment strategy is largely focused on researchers who may have a Contrarian thinking about how linguistic models are manufactured right now,” says Lewis.
