Ivo Released Sage AI Model for Legal Review

The open-source model aims to automate contract analysis by meeting internal playbook standards at a 51 percent accuracy rate.

Updated on Oct. 1, 2026 in Artificial Intelligence

Isometric editorial illustration featuring stacked rectilinear volumes and a column, representing the structural nature of automated legal contract analysis. — Ivo has launched its Sage AI model, developed with River AI, to automate routine contract reviews for enterprise legal teams using machine-readable intelligence. AI Illustration. Upload story photo >

Live Poll

Do you trust AI-driven software to handle critical legal contract reviews for your business?

Ivo has released Ivo Sage, an open-source model post-trained on DeepSeek V4 Flash to assist with enterprise legal contract reviews. Developed in partnership with River AI, the tool is currently being deployed to Fortune 500 legal teams.

Why it matters

This release seeks to standardize contract review processes by turning legal documents into machine-readable intelligence. It addresses the high manual overhead of contract editing by attempting to automate routine adjustments and compliance checks.

The Ivo Sage model achieved a 91 percent score on the Legal Agent Benchmark Contracts, though it currently meets only 23 percent of expert attorney escalation criteria. When reviewing counterparty changes, the model correctly concedes edits 82 percent of the time but only rejects changes 46 percent of the time.

The players

Ivo

A San Francisco-based developer of legal-focused artificial intelligence tools for enterprise contract management.

River AI

A collaborative partner providing technical infrastructure for the training and deployment of the Ivo Sage model.

The details

The model uses a combination of public data and synthetic text generated by attorneys to learn legal language patterns. It operates through the Ivo-micro1 Contract Bench, which evaluates effectiveness across five dimensions, including adherence to established contract playbooks. These playbooks serve as the baseline for standard positions, which the model currently adheres to at a 51 percent success rate.

Ivo released the Sage model on October 1, 2026.
The company hosted its Inscribe conference on October 1, 2026.
Public benchmark results are expected in the coming weeks.

The Tech Race

Ivo is positioning its model against general-purpose LLMs like GPT-6 Astra by utilizing the Legal Agent Benchmark to measure specific efficacy. This effort aligns with a broader industry push to establish standardized performance metrics for domain-specific professional workflows.

The tool is currently available for Fortune 500 legal teams, with workflows centered on automating contract-change acceptance and policy compliance. Users should note that while the model handles standard playbook positions, it currently requires expert human oversight for complex escalation criteria.

The takeaway

The performance of Ivo Sage shows that while automation can handle the majority of routine counterparty concessions, complex new contract points remain a challenge. Observers should monitor the upcoming public release of the Ivo-micro1 benchmark data to see if error rates in complex escalation scenarios improve.