Applied Compute Inc., a new startup by former OpenAI researchers, aims to create custom artificial intelligence models tailored for enterprise needs. Recently, it secured $80 million in funding from Venture capital firms like Benchmark and Sequoia, amid reports of a potential $500 million valuation. Led by CEO Yash Patil, who previously contributed to OpenAI’s Codex, along with Rhythm Garg and Linden Li, the company promises bespoke AI training based on client data, offering higher output quality than standard models like GPT-5. Utilizing reinforcement learning (RL) and techniques like Low-Rank Adaptation (LoRA) speeds up model customization. Already collaborating with clients like DoorDash and Cognition AI, Applied Compute is reducing AI model development time from months to mere days. As it operates a large GPU cluster, additional funding may be on the horizon to support its growth, reinforcing its position in the rapidly evolving AI landscape.
Source link
