TypeSafe AI has released its Jev decision model in beta on the Venice API, allowing developers to retrieve typed answers rather than generative text. The model processes up to 64K tokens per call and costs $0.042 per 1 million input tokens, while output tokens are provided for free. Users supply an application's state and a typed question to receive a result as a probability, a specific chosen option, or a rubric score.

The system eliminates the requirement for JSON prompting and parsing layers often used with traditional chat models to extract structured data. Jev is also available via OpenRouter and the Vercel AI Gateway for use cases such as routing support tickets and flagging fraud. On published workflows, the model is up to 190x faster and 440x cheaper than standard large language models.