⚡ Google's new Gemini Decision 1 answers up to 1,000 yes/no questions in a single pass
Google DeepMind launched the Gemini Decisions API, powered by a new model, Gemini Decision 1, built for bounded choices like tool selection, policy checks and model routing. Instead of writing text token by token, it scores the options in one pass over text, images, audio, video or PDFs, returning probabilities and confidence scores in about 150-250 ms. It costs $0.10 per million input tokens with no charge for output, and Google says it already routes queries in Gemini Notebook and guards MCP tools in Search.
⭐ Agents make thousands of small decisions, and a fast, cheap model built just for choosing could cut the cost and lag of every step.
#AI
Google DeepMind launched the Gemini Decisions API, powered by a new model, Gemini Decision 1, built for bounded choices like tool selection, policy checks and model routing. Instead of writing text token by token, it scores the options in one pass over text, images, audio, video or PDFs, returning probabilities and confidence scores in about 150-250 ms. It costs $0.10 per million input tokens with no charge for output, and Google says it already routes queries in Gemini Notebook and guards MCP tools in Search.
⭐ Agents make thousands of small decisions, and a fast, cheap model built just for choosing could cut the cost and lag of every step.
#AI
