---
title: Assistant and agent
description: Clipia Agent and Clipia Chat — models from OpenAI, Anthropic, Google and others, an agent with generation tools, attachments, memory, export and limits.
icon: Bot
---

The [AI agent](/en/agent) page brings together two helpers. **Clipia Agent** takes a task in plain words — "make a thumbnail for a travel video" — picks a model, writes the prompt and runs the generation itself. **Clipia Chat** is a regular chat with language models for writing, ideas, document analysis and questions.

<Actions>
  <Action href="/en/agent" icon="Bot" primary>Open the agent</Action>
  <Action href="/en/docs/studios" icon="LayoutGrid">Studios</Action>
</Actions>

<Callout type="info" title="Prices">
Prices are current as of October 2026. The exact cost is always shown on the run button before anything is charged.
</Callout>

## Which one to use

| Task | Where |
| --- | --- |
| Get an image or a video without learning the models | Clipia Agent |
| Pick a model and settings for an idea | Clipia Agent or the model picker in the [studios](/en/docs/studios) |
| Write copy, a script, a post or an email | Clipia Chat |
| Analyze a PDF or Word document, get a summary | Clipia Chat with an attachment |
| Find up-to-date information online | Clipia Chat with web search |

The `/assistant` address also works and opens the same page.

## Getting started

<Steps>
<Step>
### Open the page

Go to [/agent](/en/agent) and choose Clipia Agent or Clipia Chat.
</Step>

<Step>
### Pick a model

The model picker shows the developers and labels: premium models are marked PRO and show an approximate price, "≈N credits per message".
</Step>

<Step>
### Describe the task

Say what you need. You can attach a photo or a document. The more specific the task, the better the result.
</Step>

<Step>
### Keep the conversation going

Clarify, ask for changes, ask follow-up questions. The helper keeps the context of the conversation.
</Step>
</Steps>

## Models

The chat offers models from OpenAI, Anthropic, Google, xAI and DeepSeek. The lineup is updated as new versions come out — the model picker always shows the current list.

| Type | Who can use them | Cost |
| --- | --- | --- |
| Standard | Everyone | No credits charged. Without a subscription — within 100 messages a month |
| Premium (PRO label) | Any paid plan | Charged in credits, usually 1–4 per message. The approximate price is shown next to the model |

If a response fails, no credits are charged for that message.

## Response settings

| Setting | Options | Why |
| --- | --- | --- |
| Web search | On / off | The model looks up fresh information online before answering |
| Reasoning depth | Fast, Balanced, Deep | Deeper is more accurate on hard tasks but slower |
| Answer length | Brief, Normal, Long | How much text the answer contains |
| Response style | Precise, Balanced, Creative | Precise for facts and instructions, Creative for ideas and copy |

## Attachments

| Type | How many | Limits |
| --- | --- | --- |
| JPEG, PNG and WebP images | Up to 3 per message | Only for models that understand images |
| PDF, DOCX, TXT and MD documents | Up to 2 per message | Up to 20 MB and 120,000 characters of text each |

If a model does not understand images, choose another one — for example, from Google or OpenAI.

## The agent and its tools

The agent does more than answer: it calls Clipia tools. All tools are available on every plan.

<FeatureGrid cols={2}>
  <Feature icon="Image" title="Image and video generation">The agent picks a model, writes the prompt and runs the generation. The result appears in the chat and in My works.</Feature>
  <Feature icon="LayoutTemplate" title="Template search">Finds ready-made templates that match your idea.</Feature>
  <Feature icon="Cpu" title="Model catalog">Picks a model for the task and the budget.</Feature>
  <Feature icon="Wallet" title="Balance">Checks how many credits you have.</Feature>
  <Feature icon="WandSparkles" title="Prompt enhancement">Adds details to a scene description.</Feature>
  <Feature icon="History" title="Recent works">Looks at your recent generations to continue or repeat them.</Feature>
  <Feature icon="ScanEye" title="Visual review">Evaluates a finished result and suggests improvements.</Feature>
  <Feature icon="Search" title="Photo reference search">Finds openly licensed photos on Wikimedia Commons as a reference for a place, object or style.</Feature>
</FeatureGrid>

### Confirming expensive generations

- The agent runs generations of up to 100 credits on its own.
- If a generation costs more, a "This generation costs N credits" card appears with a "Confirm · N credits" button. Nothing runs until you click it.
- Credits are charged after the generation completes successfully.

## Memory

The agent remembers important facts about you — your niche, style, preferred formats — and uses them in later conversations. Manage memory on the [Agent memory](/en/account/memory) page or with the Memory button in the chat:

- view and delete individual facts;
- pause memory;
- download a memory archive;
- delete all memory at once.

## Conversations and export

- The conversation list opens in a side panel; on desktop press <Kbd>⌘</Kbd> <Kbd>B</Kbd> (<Kbd>Ctrl</Kbd> <Kbd>B</Kbd> on Windows). You can search, pin, rename, archive and delete conversations.
- Use Export to download an answer as PDF, Word (.docx), Excel (.xlsx) or Markdown (.md).
- If an answer is cut off, a "Response truncated" label and a Continue button appear.

## Limits

| | No subscription | Any paid plan |
| --- | --- | --- |
| Messages per month | 100 | No monthly quota |
| Models | Standard | Standard and premium |
| Rate | Up to 10 messages per minute | — |
| Agent tools | Up to 5 per message and 50 per day | No limit |
| Enhance prompt | Up to 60 per minute | — |

Generations the agent runs are paid for in credits at the same prices as in the studio.

## Troubleshooting

| Message | What to do |
| --- | --- |
| Monthly limit reached | The 100-message limit resets at the start of the next period. Paid plans have no monthly quota |
| Requires subscription next to a model | This is a premium model. Choose a standard one or pick a [plan](/en/tariffs) |
| Tool limit reached | The agent keeps answering without tools. The daily limit resets the next day; paid plans have no limit |
| Response truncated | Click Continue |
| The model cannot see the image | Choose a model that understands images |

## Tips

- Give the agent context: what the image is for, where it will be posted, what format you need.
- Ask for several prompt options, then run the best one.
- Double-check important facts: answers are generated by AI, and the chat reminds you with the note "You are chatting with the Clipia AI system".