← All documentation
Guide 02 — Telegram Bot

Telegram Bot Commands

Chat with Dolphin through our advanced multimodal Telegram bot.

Dolphin Bot has a few commands that can be used for RAG, Audio Generation, context history & personality settings.

/menu

/menu can be used to change the AI model's parameters between Logical & Creative mode.

You can also enable & disable context memory from the menu.

Turning context off means that every request to the bot is completely independent, meaning that the AI model will not have access to any of your previous messages.

/clearcontext

Clear context wipes the conversation history from the bot's memory.

We recommend using this regularly when changing conversation topics for the best quality responses.

/search [prompt] is the format used to start a web search.

Example:

/search what are the 10 fastest cars in the world?

The bot will:

  • Search the web & extract text from the top 5 webpages in the results
  • Feed this text into the current model you have selected
  • Generate a summary of all 5 webpages
  • Keep the extracted data in the model's context memory for future reference

/scan

/scan [URL] is the format used to start a scan.

Example:

/scan https://bitcoin.org/bitcoin.pdf
Note: Dolphin Llama3 8B Instruct currently has a limit of 2730 tokens per scan.

We will soon be releasing Dolphin Llama3.1 8B Instruct which will have 131k context & be able to perform very large scans.

The scanned text will stay in the model's context memory & can be referenced in following prompts.

/tts

/tts [prompt] is the format used to generate both text & audio replies from the current model you have selected.

Example:

/tts what is the shape of Italy?

The prompt will be sent to the LLM model you currently have selected.

The response from the model will be converted into audio using AI Text to Speech instances & sent to you alongside the original text response from the model.

/speak

/speak [phrase] is the format used to convert [phrase] from text directly into audio using our Text to Speech instances.

Example:

/speak she sells seashells on the sea shore

/voice

/voice is used to switch between different voice profiles that will be used for audio generation.