The Directory

AI Tools

New launches tracked daily — what they cost, who they're for, how to get the most out of them

All tools

Agentsby Microsoft

Decision-1

Name: Decision-1
Author: Microsoft

Free

Fast AI classification and routing model for decision-making tasks

Decision-1 is Microsoft's lightweight AI model built on Qwen3.5-9B, optimized for rapid classification and routing tasks. Achieving 83.5% accuracy with just 85ms latency across 36 benchmarks, it excels at making quick decisions in multi-agent systems, content moderation, and workflow automation. The model is designed to handle high-throughput scenarios where speed and accuracy are critical for routing requests to appropriate specialized models or systems.

Who it's for

AI engineersDevOps teamsProduct managers building multi-agent systemsBackend developersMLOps professionals

Pricing · free

checked Oct 10, 2026

Plan	Price	Includes
Free	Free	Open-source model available on Hugging Face · Self-hosted deployment · No usage limits · Commercial use permitted

AI-researched pricing — verify on the official site before subscribing.

Use it for

— Routing user queries to specialized AI models
— Content moderation and filtering
— Intent classification in chatbots
— Workflow automation decisions
— Real-time data triage
— Multi-agent orchestration

Get the most out of it

01Use Decision-1 as a lightweight router before calling more expensive specialized models to reduce costs
02Optimize inference by batching requests when latency requirements allow for it
03Fine-tune on your specific classification tasks to improve accuracy beyond the 83.5% baseline
04Deploy close to your application servers to minimize network latency and achieve sub-100ms response times
05Combine with embedding models for semantic routing in RAG applications

Visit Decision-1

Stacks featuring Decision-1

For DevSecOps lead at a Fortune 500 implementing multi-agent workflows who needs real-time oversight and kill-switch capability

Enterprise Agent Safety Monitor

Agent Native CloudVulnHunterDecision-1BuzzAX Platform

A production monitoring stack that logs every AI agent action, flags risky decisions in real time, and provides instant rollback via emergency circuit breakers. Built for enterprises deploying autonomous agents across customer-facing and internal operations.

How the workflow runs

01agent-native-cloud — Deploy and orchestrate all production AI agents on enterprise-grade infrastructure with built-in observability
02decision-1 — Route agent decisions through fast classification layer to detect high-risk actions before execution
03vulnhunter — Continuously scan agent code and decision paths for exploitable vulnerabilities or unintended behavior
04buzz — Stream all agent activity logs and alerts to centralized team dashboard for human-in-the-loop review
05ax-platform — Apply organizational AI maturity framework and enforce safety policies with mandatory approval gates

This stack directly addresses the industry-wide trust crisis and Microsoft's call for emergency brakes. It gives enterprises the confidence to deploy agents at scale while maintaining human oversight and instant intervention capability—critical as regulatory frameworks emerge and liability questions intensify.

Total (5 tools)Free

In the news

Microsoft's Decision-1 model enters the fast-growing AI decision model raceThe Decoder · Oct 10, 2026

Was this useful?