Skip to main content
Levanto Sage is a decision model for workflows where software needs to decide, not write. It turns content plus a question into a structured decision with a probability or confidence signal. It behaves like a general-purpose classifier, but one that weighs context and meaning instead of matching on similarity alone — no training needed, ready to use immediately for your use case. Use Sage when an application needs to route, approve, block, score, rank, tag, or escalate. Unlike LLMs that return text, Sage always returns a machine-actionable decision signal.

​
Supported decision kinds

Yes/No

Binary questions with probability and confidence.

Choice

Pick the best option from a set.

Scale

Rubric scoring with expectation and confidence.

Sort

Rank items by a criterion.

Tags

Apply independent yes/no labels.

Quickstart

Make your first /decide call.

​
Documentation sections

  • Use cases — Agents, operations, guardrails, moderation, and risk & fraud.
  • Decision model — HTTP API for Yes/No, Choice, Scale, Sort, and Tags decisions.
  • Authentication — API key headers and error codes for /decide calls.
  • Limits — Per-kind size limits and the content and question token budget.
  • Errors — Status codes and error response bodies.
  • Pricing — Token pricing ($3 / 1M input), grounding surcharges, and enterprise volume discounts.