Choosing the Right ASI:One Model | ASI:One Documentation

The ASI:One family offers specialized models designed for different use cases and performance requirements, each optimized for specific aspects of agentic AI and Web3 applications.

Explore Models

ASI:One Mini

ASI:One Fast

ASI:One Extended

ASI:One Agentic

ASI:One Graph

Quick Comparison

Model ID	Context Window	MMLU Benchmark	Ideal For
`asi1-mini`	32 K tokens	85%	Everyday agent workflows, chatbots, scheduling agents
`asi1-fast`	24 K tokens	87%	Ultra-low latency, real-time applications, instant agent discovery
`asi1-extended`	64 K tokens	89%	Deep reasoning, complex analysis, multi-hop retrieval
`asi1-agentic`	32 K tokens	85%	Agent discovery & orchestration, delegation planning
`asi1-graph`	32 K tokens	85%	Data visualization, chart generation, visual explanations

Note: Along with asi1-agentic, asi1-fast-agentic and asi1-extended-agentic are agentic models that can also call agents for you.

When Should I Use Each Model?

Mini — General-purpose model for everyday agentic workflows and Web3 applications
Fast — Ultra-low latency model for real-time applications and instant agent discovery
Extended — Enhanced reasoning capabilities for complex analysis and deep audits
Agentic — Built-in agent selection & orchestration capabilities
Graph — Specialized for data visualization and turning numbers into comprehensible visuals

FAQ

Are all models tool-calling capable? Yes. asi1-agentic, asi1-extended-agentic and asi1-fast-agentic are agentic models that just do the planning for you.

What does asi1-graph do? asi1-graph creates graphs for datasets.

Do I need separate API keys? No. The same key works across all models; billing is based on usage.

_{Last updated · 2025-07-30}