Auto

AI model · Language model (LLM) · Text → Text by Notion

Frontier-quality AI results at 90% lower cost.

Auto is Notion's proprietary upgraded model router designed to provide frontier-quality AI results at a significantly lower cost. It serves both individual users and businesses by matching specific tasks to the most appropriate underlying model based on quality, speed, and efficiency requirements. The system is capable of changing course if it detects a model is stuck, ensuring reliable task completion for Notion Agents and Custom Agents. Notion reports that Auto delivers results comparable to leading models like Opus 5.5 and Astra while reducing average costs by approximately 90%. It is fully

Model specs

The full spec card (context, pricing per token, modalities and more) on AntSpace. Sign up free to see the full spec card.

Launched
2026-10-19

Category: Machine Learning

License
Proprietary
Access
Built into Notion AI, Notion Agent, and Custom Agents
Released
2026-10-19
Cost Reduction
90% lower average cost
Task Coverage
20+ common tasks
Credit Efficiency
Credits go approximately 10x further
  • 90% lower average cost than major frontier models
  • Frontier-quality results similar to Opus 5.5 and Astra
  • Matches each task to the right model for speed and quality
  • Automatically changes course if a model gets stuck

Overview

Auto is Notion's proprietary upgraded model router designed to provide frontier-quality AI results at a significantly lower cost. It serves both individual users and businesses by matching specific tasks to the most appropriate underlying model based on quality, speed, and efficiency requirements. The system is capable of changing course if it detects a model is stuck, ensuring reliable task completion for Notion Agents and Custom Agents. Notion reports that Auto delivers results comparable to leading models like Opus 5.5 and Astra while reducing average costs by approximately 90%. It is fully integrated into the Notion AI workspace and managed through a unified credit system.

Model Orchestration

  • Intelligent Model Matching: The router automatically matches each task to the most appropriate model by balancing quality, speed, and cost requirements.
  • Self-Correction Capabilities: If the initial model gets stuck on a task, the system automatically changes course to find a solution.
  • Frontier-Quality Results: Auto delivers results comparable to leading frontier models like Opus 5.5 and Astra across more than 20 common tasks.
  • Cost Optimization: The system provides high-quality AI results at an average cost that is approximately 90% lower than using major frontier models directly.

Enterprise Controls

  • Workspace Integration Integration: The router is integrated into the Notion AI workspace to power Notion Agents and Custom Agents.
  • Unified Credit System: Usage of Auto is managed through a unified credit system that brings together all AI usage in one place.
  • Governance and Analytics: Admins can monitor usage by person, feature, and agent through the AI Command Center.

Benchmarks

  • Quality: Similar to Opus 5.5 and Astra across 20+ common tasks

Use cases

  • Teams use Auto to process recurring tasks through Custom Agents while reducing their total AI credit consumption by 90%.
  • Project managers rely on Auto to handle complex reasoning tasks with quality similar to Opus 5.5 without manual model selection.
  • Developers build automated workflows that use Auto to ensure tasks are completed even if an initial model gets stuck.

Specifications

License
Proprietary
Access
Built into Notion AI, Notion Agent, and Custom Agents
Released
2026-10-19
Cost Reduction
90% lower average cost
Task Coverage
20+ common tasks
Credit Efficiency
Credits go approximately 10x further
Performance Target
Opus 5.5 and Astra quality levels

How it works

  1. Task Assignment: A user assigns a task to a Notion Agent or Custom Agent within the workspace.
  2. Intelligent Routing: The Auto router analyzes the task and selects the most appropriate model based on the required quality, speed, and cost.
  3. Execution and Recovery: The selected model performs the work; if it encounters an issue, Auto automatically switches to a different model to complete the task.
  4. Result Delivery and Tracking: The final result is delivered to the user, with usage tracked against the workspace's Notion AI credit balance.

FAQ

How does Auto's performance compare to other models?

Across more than 20 common tasks, Auto delivers results similar to leading models like Opus 5.5 and Astra.

What happens if Auto gets stuck on a task?

If a model gets stuck during a task, Auto is designed to automatically change course to ensure the work is completed.

How does Auto affect my Notion AI credits?

Auto usage is managed through Notion's new credit system, where your credits go approximately 10x further due to the 90% lower average cost of the router.

Added by AntSpace from public sources (notion.com).

More AI models by Notion