Last updated 08-13-2026
Category:
Reviews:
Join thousands of AI enthusiasts in the World of AI!
TavernAI
TavernAI lets you run AI roleplay chats locally on Windows or Linux with full control over characters, prompts, and story branches. Version 2.3 ships as a portable 300 MB download that stores every conversation on your machine and connects to your own model provider or loads GGUF language and vision models with CPU, CUDA, or Vulkan.
Where most character chat apps lock you into one cloud API and a single persona per thread, TavernAI 2 treats each character as a full prompt system inside a card. You get branching swipes, multi-generation replies, dynamic cast changes mid-scene, and a unified Prompt Manager that assembles exactly what the model sees before each request.
The trade-off is setup depth. TavernAI is built for people who want control over prompts, macros, and PM Scripts that turn chats into playable AI games, not a plug-and-play consumer chatbot. You bring the model costs and handle installation, including optional VPS hosting on port 8800.
It fits roleplay communities, fiction writers testing dialogue, and self-hosters who want offline-capable companion chats without surrendering character data to a hosted service. The portable package needs no installer and keeps user data in a local user_data folder beside the app.
Portable 300 MB app for Windows 10/11 and Linux with no installer or account required
Loads GGUF language and vision models locally via CPU, CUDA, or Vulkan, or connects to built-in API providers
Branching swipes let any message split into alternate reply paths that continue independently
Each character card carries its own Prompt Manager with files, images, rules, and token counters
PM Scripts turn chats into interactive AI games with custom panels and persistent scene state
VPS install script deploys to ~/TavernAI and serves the UI on port 8800 through a browser
Completely free app with local data storage on your machine
Deep Prompt Manager with file attachments, macros, and token inspection
Runs local GGUF models or connects to external API providers
Branching chats and multi-generation parallel replies
VPS deployment script with systemd auto-start support
Requires manual provider setup and model configuration
300 MB download plus model files can consume significant disk space
Steep learning curve compared to hosted character chat apps
VPS exposure on port 8800 needs password protection and firewall rules
Is TavernAI free to download?
Yes. TavernAI is a free portable app published on GitHub Releases for Windows and Linux. You pay only for the AI model provider or hardware you choose to run.
What platforms does TavernAI support?
TavernAI runs on Windows 10 and 11 as a portable zip and on Linux as a tar.gz archive. A curl installer also deploys TavernAI on a Linux VPS accessible at port 8800.
Can TavernAI run models locally?
Yes. TavernAI can load GGUF language and vision models directly inside the app using CPU, CUDA, or Vulkan. You can also switch the same chat to external API providers.
How does branching work in TavernAI?
TavernAI lets you swipe any message to create an alternate branch in the same chat. Each branch can continue as its own storyline, and multiple AI generations can run in parallel.
Where does TavernAI store chat data?
TavernAI keeps all chats, cards, and imported files in a user_data folder inside the app directory on your machine. Nothing is sent to a TavernAI cloud account.
Can I use TavernAI on my phone?
TavernAI opens in a browser, so a phone on the same Wi-Fi network can connect to a PC or VPS running the app at http://IP_ADDRESS:8800.