Decision models differ from traditional large language models (LLMs) in that they specialize in selecting from predefined answers rather than generating text and code. This can mean choosing which tool an AI agent should use next, identifying relevant files, or determining whether a test has passed. Their primary attractions are speed and cost: these decisions occur frequently during an agent's execution, and delegating them to smaller, specialized models reduces reliance on expensive frontier LLMs.
In a blog post published on Monday, Subquadratic says Snap can process 100,000 tokens and return a decision in approximately one second, potentially allowing AI agents to make decisions across large codebases, documents, and interaction histories without breaking information into smaller chunks.
Subquadratic CEO and co-founder Alex Whedon acknowledged in an email interview that Jev's arrival helped shape the company's own entry into the market. "Snap was developed with inspiration from Jev, because decision models are highly thesis-aligned," Whedon said. "Decision models remove the need for decoding for some problems."
However, Whedon believes the technology has much greater potential than the supporting roles decision models currently perform, such as evaluating LLM outputs or selecting which model should handle a task. "We see a vacuum of vision for decision models," he said. "Decision models become exciting when they actually take work away from frontier LLMs."