The AI Weekly newsletter reports that Google has released its own decision model, following similar releases from OpenAI and Microsoft in October. TypeSafe AI released the first one, called Jev, on September 15.

The source describes decision models as a kind of AI model that does not write text. A user supplies an input and a fixed list of answers. The model returns the best answer along with a probability, in milliseconds.

According to the report, Google, OpenAI, Microsoft and TypeSafe AI each charge $0.10 or less per million input tokens, and none charge for output.

The newsletter also says Shanghai AI Lab claims its 4B model beat Jev on its tests. It further states that Microsoft built its model on Alibaba's Qwen. The report says chatbots still score a little higher on the hardest decisions, but at 15 times the cost.

The newsletter adds that on Vercel's gateway, decision models handled one in four AI requests in October. This figure comes from the source and has not been independently verified.