# Usage Limits
### Chat credits (daily cap)
* **Allocation:** You receive **500 credits per day** for AI chat. The header shows **remaining / 500** (for example `438 / 500`).
* **What spends credits:** Sending messages in chat—including follow-ups, answers to multiple-choice questions, and Atlas’s replies. Cost grows with **how long and complex** each exchange is (longer text, more steps, deeper research-style answers).
* **What does not spend chat credits:** Browsing the dashboard, opening and managing strategy cards, portfolio tracking, settings, and **running backtests** on your own strategies. Those stay outside the chat credit meter. (See Credits for the full breakdown and rough cost examples.)
**Reset:** Credits refill on a **daily schedule** at **midnight UTC**. Unused credits **do not** roll over—you start again at 500 after each reset.
**Running out:** When you hit **0**, chat is blocked until the next reset. The UI shows a message such as **“Credits exhausted. Resets in Xh Ym.”** with a countdown. You can still use the rest of Atlas while you wait.
When your balance drops **below 50**, the counter turns **red** as a heads-up.
### Switching models (Atlas Lite vs Atlas)
Below the chat input you can choose:
| Model | Typical use | Credits |
| ------------------------ | --------------------------------------------------- | --------------------------------- |
| **Atlas Lite (Default)** | Fast answers, straightforward strategy searches | **3x-4x** lower cost per exchange |
| **Atlas** | Harder questions, comparisons, multi-step reasoning | **Higher** cost per exchange |
You may switch **at any time between messages**. If you are near the end of your daily credits, **Lite** is the safer default for simple searches; switch to **Atlas** when you need extra depth and accept that it will usually burn credits faster.
More context: Chatting with Atlas (model selector and sessions).
### Saving credits when a chat gets long
Atlas uses your **current session’s history** so follow-ups make sense—but **every new message can include the whole thread** when Atlas replies. As a chat grows very long, each send tends to **cost more credits** than the same question asked in a **short** thread, because there is more context to read and summarize.
**Practical habit:** When you move on to a **new topic or a new search** and the sidebar thread is already huge, click **New Chat** instead of continuing forever in the same session. You keep older chats in the list if you need them; you are not deleting your account—just starting a **clean context** so the next questions do not drag an unrelated novel of messages into every reply.
Combine that with **specific first messages** (asset, timeframe, indicator, goal) so Atlas needs fewer clarification rounds—see Prompting tips and Finding strategies.
{% hint style="info" %}
**Quick checklist:** (1) Prefer **Atlas Lite** for simple jobs. (2) Use **New Chat** for unrelated questions after a long thread. (3) Put the important constraints in **message one**.
{% endhint %}
### Other limits
Sometimes Atlas shows a short message if something is busy, unavailable, or needs a retry. Follow any on-screen instruction (“wait,” “try again,” and so on). If nothing helps after a short wait, use AlgoAlpha Support & Knowledge.
---
# Agent Instructions: Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.
Perform an HTTP GET request on the current page URL with the `ask` query parameter:
```
GET https://algoalpha.gitbook.io/algoalpha-user-manual/atlas-ai-backtesting-agent/usage-limits.md?ask=
```
The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.
Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.
