GitHub radar

An anime companion with voice and memory that "lives" beside you

OpenAmika is an open-source anime girl in Live2D (an animated flat picture) who talks with you by voice and text in real time and remembers what you told her. Install it and you get a working companion on your own PC, not a demo.

01zeekk0/OpenAmika 454Python

A desktop app with an anime character who talks by voice and text and remembers your conversations. Her body, ears and voice run on your PC; the language model that writes her replies comes from the cloud or runs locally.

When it helps

Useful if you want a standing AI companion just for company and casual chat. Skip it if you're on a Mac, a laptop without an NVIDIA GPU, or looking for a serious AI assistant for work.

Pros

  • Code is free, MIT license, reuse it however you like
  • Remembers past conversations instead of starting fresh each time
  • Cloud option via OpenRouter costs $1.50–5/month, no beefy GPU needed
  • Browser version runs with no install at mekahime.com

Cons

  • Needs Windows and an NVIDIA RTX 20-series or newer card with 8–16 GB VRAM
  • Running fully local without OpenRouter needs a 27B to 70B parameter model
  • Her voice, Live2D model and art aren't MIT, personal-use license only

How to set it up — step by step

  1. 1Open an AI coding agent (Claude Code, Codex) in your project folder on Windows with an NVIDIA GPU.
  2. 2Send the agent the text from the block below.
  3. 3For the language model, give the agent an OpenRouter key or ask it to set up Ollama locally.
  4. 4Wait for setup to finish (about 45 GB on disk) and let it launch.
  5. 5Check how the companion replies by voice and text, and ask her a couple of questions.

Text for your agent

Copy this and send it to your agent — Claude Code, Codex, any of them:

Set up the anime companion from MekaHime's GitHub here: https://github.com/zeekk0/OpenAmika — follow the README on Windows with an NVIDIA GPU, use an OpenRouter key for the language model (ask me for it) or set up Ollama locally, run it, and show me how she responds.

You need Windows 10 or 11 and an NVIDIA RTX 20-series or newer GPU with 8 GB of VRAM (16 GB if you turn on her camera eyes); setup takes about 45 GB of disk space. Not an option for a Mac or a laptop without a discrete GPU.

Open on GitHub