TypeSafe AI launches Jev, an AI decision-making model

TypeSafe AI has launched Jev, an artificial intelligence model designed specifically for decision-making. This model, named after economist William Stanley Jevons, aims to provide quick and cost-effective solutions. TypeSafe AI, co-founded by Diogo Almeida, a former member of OpenAI, has created this new model known as System One, which has certain limitations in its usage but stands out for its affordability and high speed.

Jev accepts questions in natural language similar to large language models (LLMs) but restricts its responses to choices, scoring, and agreement with a score ranging from 0 to 1. This limitation on response format means that Jev cannot be utilized as flexibly as LLMs. However, it significantly excels in speed, answering questions within 70 to 500 milliseconds, allowing for near-instantaneous operation. Additionally, the input cost is only 0.0042 dollars per one million tokens, making it hundreds of times cheaper than traditional LLM models.

The applications of Jev include serving as a guardrail for other AI tasks, making complex decisions that require language interpretation, and processing large data sets. While similar tasks could be accomplished using existing LLMs, Jev offers a much cheaper solution, often by tens or even hundreds of times. Currently, Jev is available only in a closed beta, requiring users to sign up in advance for access. Blognone reported the story.