Images and video right in chat

1Three ways to get an image

• By asking in chat. Write “draw a coffee shop logo in a minimalist style”: the assistant picks a model itself and briefly explains the choice (“best at text”, “can edit”, etc.).
• “Image” mode. The “+” menu next to the field → Image. The field switches to generation mode (the hint “Describe the image you want to generate…”).
• In Studio. It has more settings: Studio in the side panel.

2“Image” mode: its own settings

1. Turn on the mode

   Open “+” → Image. An “Image” panel appears above the field. To leave, close the panel with the cross or choose “+” → Image again.

2. Set the parameters

   On the panel: Size, Count, Quality (Auto, Low, Medium, High), Style (or “No style”), Change model and Improve prompt with AI. The price with the ⚡ icon is shown next to them.

3. Describe the shot and send

   You can attach reference pictures (they go as references). A shortcut is to write /image and the text of your request in the field.

3When the assistant draws on its own

The assistant draws only on an explicit request (“draw”, “make a picture”, “make a logo”). It doesn't add “illustrations for decoration”. If you ask “can you draw 7 frames?”, it replies with a plan and asks whether to launch: a paid launch begins on a command, not on a question about capability.

The model is chosen for the task: for lettering, one that works well with text; for editing a previous image, one that can edit. The assistant turns on expensive engines, as well as 2K and 4K, only if you yourself ask for maximum quality, printing or a large size. If you want the cheapest, write “with the cheapest model”; if you want a specific one, name it.

4Image batch: plan and price before launch

Write, for example: “make 5 images: a ginger cat by the window, in the rain, at night, in winter, at sunset”. You can list your own prompts and the assistant won't rewrite them. No more than 10 images per message; if you ask for more, the first ones are taken.

1. Get an “Images: N” card

   It holds each prompt, model and price. For a small and inexpensive batch (fewer than 4 images), drawing may start right away. For 4 or more images, an expensive batch and a batch with a persona, launch waits for your Create N button (for example, “Create 5”).

2. Change the plan if needed

   The Edit button opens the Model, Format and Persona fields (there is a “No persona” option).

3. Follow progress

   Cells go through the statuses “Queued” → “Drawing” → done. Up to three images are drawn at once (the account's overall generation limit). At the top you see “Done K of N”. You can close the tab: “You can close the chat: the images will finish and stay here”.

4. Stop or collect the result

   Stop queue interrupts unfinished cells; Download all collects the finished ones. After completion, “Charged ⚡N” appears at the bottom.

If the balance isn't enough for the whole batch, the card says “Balance covers k of N” and offers Top up; you can create as many as the balance allows. If a cell fails, it gets a Retry button; if moderation rejects it, the cell says “The model rejected the request, no charge was made”.

To compare models: “draw this on five different models” or “on FLUX and Grok”. The assistant takes your prompt unchanged and draws it on several models; in the plan you can Add model or Remove an extra one. Some models may have a “plan required …” or “temporarily unavailable” mark.

5Editing by frame number

All cells of the batch are numbered. Write “in image 3 make the night sky” or “make the third one brighter”: a new cell “edit #3” appears and the original stays. The Another variant button on a cell creates “variant #N”. If you need a different frame from the batch, just name its number.

Clicking a cell opens a viewer: arrows to move between frames, Fit for scaling, Download this one. The field “What to change in this image…” and the Edit this image button send an edit for the selected frame.

6Hero: a persona from your materials

If a request includes a name (“with Alex”, “with me”), the assistant looks for your persona: a saved face, object or place from Studio. The name can be written any way: in Cyrillic or Latin letters, in any grammatical case. If two personas fit (“Alex” and “Alex 2”), it asks one question with buttons and doesn't guess. That question has the options “No persona” and “Create persona”.

A batch with a persona always waits for the launch button. How to create a persona is on the page about storyboards and persona from chat. You confirm the rights to the face yourself.

The same image batch works in the “Models” section and in a chat with a soul (agent). There is no batch in agent groups.

7Video from chat and a video batch

1. Ask for a video

   For example: “Make a video: a cat barista makes a latte, 5 seconds, vertical”. If you didn't specify details, the assistant uses the model's default parameters without interrogating you.

2. Get a “Video” card

   It holds the model, mode, duration, format and the exact price, and at the bottom “⚡N will remain”. Until you click, nothing is launched or charged.

3. Launch or change

   Generate launches the video. Edit sends a request to adjust parameters (“make it 10 seconds”, “vertical”), and the assistant sends a new card.

4. Wait for the result

   The card will say “Generating video…”; you can close the chat. The result will appear here and in Studio → My generations. The Stop button interrupts the task; if it hasn't started yet, no tokens are charged.

For several videos: “make 3 clips: …” or “animate images 2 and 5”. This is a video batch: confirmation is always required with the Create N button; no more than 4 videos at a time. Videos are sent to work one by one, and assembly takes a few minutes: “Videos take a few minutes to assemble; you can wait here or in Studio”.

8What chat doesn't have

• Agent group chats have no image batch, video batch, storyboard or persona from chat.
• The video card (a single video and a video batch) is in the main chat. In the “Models” section and in a chat with a soul there is an image batch, a storyboard and a persona, but no video card; for video, open the main chat or Studio.
• Music and voiceover are not created from chat; they are made in Studio.
• Mobile apps receive these cards as they update; if a card isn't there, use the web version.

Related