What is Auto on AI Plaza?

Auto (Smart) selects a capable model under cost caps at send time—good enough for the ask, not always the cheapest micro model. It uses message content, length, and tool context. You can still lock any available model manually.

What does the usage percent mean?

The meter shows how much of your plan’s fair-use allowance is used. Marketing figures on Pricing are illustrative capacity indexes—not a promise of identical third-party API spend. See Terms.

Getting started · 9 min read

Models, usage meters, and Auto on AI Plaza

How AI Plaza Auto smart routing works, how usage percent is calculated, and how to stretch Pro, Business, and Enterprise fair-use capacity.

Published 2026-07-11 · Updated 2026-07-12 · AI Plaza

Models, usage meters, and Auto on AI Plaza

Understanding Auto and the usage % meter helps you get quality without burning the plan.

Auto (Smart) routing

When the model picker is set to Auto, AI Plaza:

Reads your message (and tool context)
Estimates difficulty / modality needs (structure + precision signals; ambiguous asks lean up one tier)
Picks a model that is good enough under the price cap (quality floor within band—not always the cheapest micro)
May prefer Qwen / DeepSeek for Chinese-heavy text when configured

Privacy-safe routing features may be logged to improve Auto over time. You can override Auto anytime for a specific reply.

Usage percent (Plan & usage)

Registered plans show a Total % bar on Account and in Chat:

Economy models move the bar slowly
Flagship models move it faster
Tools still consume usage based on the underlying model tokens

Fair-use capacity is a metered administrative allowance (prompts + plan features), not a cash balance or fixed request count. Details: Terms and Plans & billing.

Practical playbook

Outline with Auto / everyday models
Use a scenario tool instead of a huge custom system prompt
One flagship pass for final polish
Export from Chat instead of regenerating for each stakeholder
Lock a flagship only when the task is hard (long reasoning, critical external docs)

Context meter in Chat

When a thread grows long, AI Plaza shows context usage. Near the limit, summarize & continue or start a new chat so quality stays high.

Try it

Open Chat, leave Auto on, and watch usage % after a few sends. Related: Getting started · Prompt engineering basics.

Frequently asked questions

What is Auto on AI Plaza?: Auto (Smart) selects a capable model under cost caps at send time—good enough for the ask, not always the cheapest micro model. It uses message content, length, and tool context. You can still lock any available model manually.
What does the usage percent mean?: The meter shows how much of your plan’s fair-use allowance is used. Marketing figures on Pricing are illustrative capacity indexes—not a promise of identical third-party API spend. See Terms.