Create AI images, video and music in our Telegram bot

You can now create AI images, video and music in the official API Stock Telegram bot. Choose a model, enter your idea, check the price and confirm the request; the result comes back to your private chat. Start directly in Telegram or connect your website account to use its existing balance.

What can you create in the Telegram bot?

The bot brings the media catalog into a conversation. You choose an image, video or music model and fill in its available settings using menus and prompts. You do not need to write an API request, copy an API key into Telegram or build a workflow before trying an idea.

That makes it useful for the small creative jobs that happen between other tasks: drafting a banner, testing a visual for a social post, making a short clip or exploring background music. You can start from a description, or supply reference media where the selected model supports it. Different models accept different inputs, so choose the model for the task rather than expecting every menu to contain the same fields.

The model catalog remains the place to compare capabilities in detail. The bot is another way to submit media requests through your account. LLM chat is not available in this launch: the supported workflows are image, video and music generation.

How do you start your first generation?

Open @api_stock_bot in a private chat and send /start. You can create an account directly through Telegram without an email address or connect an account you already use on the website. Choose the option that fits how you want to manage your balance.

Then use this sequence:

1. Send /models and select a media model.
2. Enter your prompt and complete the model's required settings.
3. Add reference media if the task needs it and the model supports it.
4. Review the quoted price and confirm the generation.
5. Wait for the result in the same chat.

For a first test, pick a simple task whose result is easy to judge. An image is a good starting point because you can inspect composition, details and any requested text before committing to a more involved video concept. Decide the intended aspect ratio first: a square product visual and a wide article illustration need different framing.

Here is a prompt you can adapt for a text-to-image model:

A wide editorial illustration of a small independent bookshop at dusk, warm light through the windows, rain on the pavement, one bicycle beside the entrance, clean composition with empty space on the left for a headline, no visible lettering, 16:9.

Choose 16:9 in the model settings when that option exists. The prompt describes the scene; the setting specifies the output format. Check whether the empty area is actually usable, whether the bicycle looks plausible and whether the model added unintended lettering. If the composition misses the brief, change one instruction before the next paid attempt.

Can you use your existing website balance?

Yes. To connect an existing account, sign in on the website and open Telegram connections. Create a connection link, open it in your own Telegram account and check the account details shown by the bot before approving. This lets the bot use the balance of that account.

Connect before making your first paid request if you want to use funds already on the website. Creating a new account in Telegram is a separate onboarding choice; do not assume that a new account will find your existing website account automatically. The explicit connection flow is how you choose which account the bot should use.

Send /account to view the connected account's balance or start a top-up. Choose a pack or enter an amount, select an available payment method and check the exact charge; the Pay button opens the hosted checkout. Refresh your balance after payment. The dashboard action provides a separate one-time sign-in link, so you can open your account without copying a login code. Treat that link as private and open it yourself.

You can manage or revoke a connection in the dashboard. If you created the account directly in Telegram, keep another sign-in method before removing its Telegram access. This matters when you later want to change how you sign in without losing your route back to the account.

How much does a generation cost?

The model and selected settings determine the price. The bot shows a quote before you confirm the debit, so review it for each request, especially when changing resolution, duration or the model's action. There is no universal price for an image, a video or a piece of music.

These examples come from the live catalog on 14 October 2026. They describe the specified catalog options, not every possible configuration:

Check the current pricing page and the bot's confirmation screen before ordering. A model can have several paid actions with different prices; the Suno music example is not the price of every operation available for that model.

A practical way to control spend is to write the brief before submitting, make one test and inspect it. For an image, check composition and reference fidelity. For a clip, check the motion and continuity. For music, listen alongside the voiceover or scene it will accompany. Each new confirmed generation is a new paid request, so use the first result to decide what to change.

How do references and results work?

When a model accepts references, you can send supported media attachments or HTTPS URLs in the field the bot asks you to fill. For a list of references, send the files individually and finish with /done. Use /skip to clear an optional field when the conversation offers that choice.

Bot uploads are limited to 20 MiB per file, with up to 20 items in a media list; a selected model may accept fewer references. These are upload limits, not a promise that every model can process every file type. Read the field prompt and prepare a smaller source file if necessary. Use media you have permission to upload and process.

Generation runs asynchronously. You confirm the request, then the bot delivers the result after processing finishes. Send /tasks to check pending requests; you do not need to submit the same idea again because a result has not arrived immediately. Your dashboard is also available for reviewing generation history.

For a more complete workflow, the guides explain individual creative tasks, while the model pages document supported inputs. Use those references when a project needs several connected steps rather than a single quick draft.

Which commands should you remember?

/models starts model selection, /account opens account actions, and /tasks checks pending work. /cancel exits the active conversation when you want to change direction. /language changes the interface language; this storefront's bot supports English, Chinese, Japanese, Korean, Russian, French, German, Indonesian, Portuguese and Spanish.

The bot operates in private chats. Start there even if you discovered it through a group or the channel announcement. For automated workflows in your own application, the API documentation remains the next step; the Telegram bot is the convenient place to try a media idea yourself.

FAQ

Do I need to register on the website first?

No. You can create an account directly through /start in Telegram without an email address. If you already have a funded website account, connect that account instead.

Do the website and bot use the same balance?

They do when the bot is connected to that website account. Check the account shown during connection approval, then use /account to view its balance.

Can I use the bot for text conversations with LLMs?

LLM chat is not part of this launch. The bot currently supports image, video and music generation through the media catalog.

Where do I top up?

Open /account, choose a pack or enter an amount and select an available payment method. Check the quoted charge, then use Pay to open the hosted checkout.

What should I do while a result is processing?

Use /tasks to check pending requests and wait for delivery in the chat. Avoid confirming a second paid request just to check the first one.