Enter a personality description. The system prompt will be: "Assume the personality of [your text]. Roleplay and never break character."
When enabled, removes the "keep your responses relatively short" guideline from the personality system prompt.
Choose your preferred color scheme for the AI Assistant.
A preview of your selected theme colors.
Enable or disable text-to-speech for AI responses
Automatically play TTS for new messages and play them in sequence
Select which voice to use for speech generation
Custom instructions for how the AI should speak. If left empty and using personality, will use the personality for instructions.
Controls randomness: Lower values make output more focused, higher values more diverse
Decreases likelihood of repeating the same phrases (-2.0 to 2.0)
Increases likelihood of introducing new topics (-2.0 to 2.0)
Maximum number of previous messages to include in context
Store your API keys securely in your browser. Keys are saved in your browser's local storage and are never sent to our servers.
Required for OpenAI models. Get API Key
Required for Grok models. Get API Key
Required for Claude models. Get API Key
Required for Gemini models. Get API Key
Required for Mistral models. Get API Key
Required for Hugging Face models. Get API Key
Required for GitHub Models. Get API Key
The URL of your Ollama server. Default is http://localhost:11434 for local instances.
The base URL of your LM Studio server. Default is http://localhost:1234 for local instances.
Enable or disable function/tool calling. Disabling this will make the assistant behave like a standard chatbot.
Toggle individual tools on/off. Changes take effect immediately.
Configure additional API keys required by specific tools. OpenAI, xAI, and Google API keys are managed in the main API Keys tab.
Required for social media, entertainment (IMDB, Steam), and news search tools. Get API Key
Required for the weather tool. Get API Key
When enabled, the assistant can store brief user-provided details via a dedicated function and include them in the system prompt.
Maximum number of short memories to retain (default 25). Oldest memories are dropped when the limit is exceeded.
Memories are captured automatically on request. Keep manual entries brief — a few words to a sentence is best. Maximum 600 characters (about three long sentences).
Location awareness allows the AI to understand your geographical context, including your timezone and general location. This information is processed locally and only shared with the AI during conversations to provide more relevant responses.
Allow the AI to know your approximate location. This adds context about your location and timezone to conversations.
Your location information is processed locally and only shared with the AI during conversations.
Version
Tyumi is an open source AI assistant platform that seamlessly integrates with multiple AI providers including OpenAI, Anthropic, Google, Mistral, xAI, Hugging Face, GitHub Models, and local Ollama models. It offers advanced features like personality customization, tool calling (web search, image generation, etc.), text-to-speech, theme customization, and secure local storage of conversations and API keys.
All of your conversations and generated images are stored locally in your browser. You can easily manage your chat history and view your images in the image gallery. Make sure to download the images you want to keep, as they are not stored on any server and will be lost if you clear your browser data.
For more information, visit the GitHub Repository.
If you find this project helpful, consider supporting its development: