KoboldAI/KoboldAI-Client

For GGUF support, see KoboldCPP: https://github.com/LostRuins/koboldcpp

KoboldAI – Browser‑based front‑end for AI‑assisted writing

What it is

  • A web UI that lets you interact with a variety of local or remote language models for story writing, text‑adventure games, or chatbot‑style conversations.
  • The repository contains the legacy client (no longer maintained for the newest models; newer work has moved to KoboldCpp).

How it works

  • The UI runs in your browser and talks to a model backend you choose: a local GPU/CPU instance, a Google Colab notebook (TPU or GPU), or an online API (OpenAI, Inferkit, etc.).
  • It provides tools such as Memory, Author’s Note, World Info, save/load, and adjustable generation settings.
  • Three main interaction modes:
    1. Novel mode – optimized for prose writing.
    2. Adventure mode – behaves like a text‑adventure (second‑person “You” commands).
    3. Chatbot mode – formats dialogue with explicit speaker tags.

Supported models

  • The README lists dozens of Hugging‑Face models that can be loaded, grouped by style:
    • Novel (e.g., Nerys, Janeway, Picard)
    • Adventure (e.g., Adventure, Skein, Spring Dragon)
    • NSFW (e.g., Erebus, Lit)
    • Hybrid/Generic (e.g., GPT‑J‑6B, OPT, NeoX, MythoMax, Chronos)
  • Different models are recommended for different modes; adventure models expect the word You, novel models do not.

Running it

  • Local: install the client, point it at a model file or an API endpoint.
  • Google Colab: two ready‑made notebooks (TPU and GPU) let you launch a VM, download a model, and connect the UI. Saves can be synced to Google Drive.
  • API: supply an OpenAI or Inferkit key to use their hosted models.

Why it matters

  • Gives non‑technical writers a way to experiment with large language models without writing code.
  • Supports both creative writing and interactive storytelling, bridging the gap between pure text generation and game‑style play.
  • Although this repo is now legacy, it still illustrates the typical architecture of a self‑hosted LLM front‑end.

Current status

  • Marked as legacy; the developers recommend switching to KoboldCpp for modern model support.
  • No active maintenance, but the code remains functional for the older models listed.

TL;DR: KoboldAI‑Client is a browser UI for running various language models (local, Colab, or API) as a writing assistant, text‑adventure engine, or chatbot. It ships with many community‑curated models and offers tools for story management, but the project is now deprecated in favor of KoboldCpp.

Related

  • Project
    huggingface/smollm

    SmollM is Hugging Face’s open‑source family of tiny yet powerful models: SmolLM‑3 (a 3 B‑parameter, 128 k‑token, multilingual instruction model) and SmolVLM (a compact vision‑language model). All weights, data mixes, and training configs are public, and the models can be loaded with the standard `transformers` library for CPU or modest‑GPU inference.

  • Project
    ilysenko/codex-desktop-linux

    A community-maintained Linux distribution of the official ChatGPT desktop app that adds native package support and a wide range of optional system-integration features.

  • Project
    yaojingang/GEOFlow

    GEOFlow is an open‑source Laravel platform that unifies knowledge‑base management, AI‑generated SEO content, automated quality checks, human review, multi‑site publishing and analytics for enterprise websites. It runs on PHP 8.3+, PostgreSQL (pgvector), Redis and Docker, and is released under AGPL‑3.0 with optional commercial licensing.

  • Project
    ML-GSAI/LLaDA

    LLaDA is a large-scale masked diffusion language model that challenges the dominance of autoregressive architectures, achieving performance comparable to LLaMA3 8B.