About
GPT Author is an open-source AI pipeline that automates the creation of full-length fantasy novels. Built as a Jupyter notebook (with both a GPT-4 variant and a newer Claude 3 variant), it chains together large language models and Stable Diffusion to plan, write, and package an entire book from a single user-supplied prompt. The workflow is straightforward: the user provides an initial story concept and specifies how many chapters they want. The AI handles everything else — outlining the plot, writing each chapter, maintaining narrative consistency, and generating cover imagery — before bundling the result into an EPUB file compatible with Kindle, Apple Books, and other e-readers. The newer Claude_Author.ipynb notebook leverages Anthropic's Claude 3 model for noticeably better prose quality with simpler underlying code. The original gpt_author.ipynb and gpt_author_v2.ipynb notebooks use OpenAI's GPT-4 with Stable Diffusion for cover art. Because it runs on API calls rather than local GPU inference, setup is minimal — open the notebook in Google Colab or a local Jupyter environment, supply your API keys, and run. A 15-chapter novel costs roughly $4 and completes in minutes. GPT Author is ideal for writers seeking rapid first drafts, worldbuilders who want AI-assisted long-form storytelling, developers exploring multi-model pipelines, and hobbyists who want to experiment with AI-generated fiction. The MIT-licensed codebase with 2.5 k GitHub stars makes it easy to fork and extend.
Key Features
- End-to-End Novel Generation: Transforms a single user prompt into a fully structured, multi-chapter novel complete with plot outline and consistent narrative arc.
- Claude 3 & GPT-4 Support: Includes separate notebooks for Anthropic's Claude 3 (higher prose quality, simpler code) and OpenAI's GPT-4 so users can choose their preferred model.
- Stable Diffusion Cover Art: Automatically generates cover imagery using Stable Diffusion to complement the written content.
- EPUB Export: Outputs a properly formatted EPUB file ready for immediate use on Kindle, Apple Books, Kobo, and other e-reader applications.
- Low-Cost & Fast: A 15-chapter novel can be generated in just a few minutes for approximately $4 in API usage costs.
Use Cases
- Generating rapid first-draft fantasy novels for writers who want an AI-assisted starting point to refine and edit.
- Worldbuilders and game designers creating lore-rich fictional narratives for tabletop RPGs or interactive fiction projects.
- Developers and AI researchers exploring multi-model orchestration pipelines combining LLMs with image generation.
- Hobbyists and enthusiasts who want to experiment with AI-generated long-form storytelling at low cost.
- Content creators producing serialized AI fiction for personal blogs, newsletters, or self-publishing platforms.
Pros
- Truly End-to-End: Handles everything from plot planning to final EPUB packaging, requiring minimal user effort beyond the initial prompt.
- Open Source & Extensible: MIT-licensed code on GitHub makes it easy to fork, customize, or integrate into larger creative pipelines.
- Multiple Model Options: Supports both OpenAI GPT-4 and Anthropic Claude 3, letting users pick the best model for their quality or cost requirements.
Cons
- Requires Multiple API Keys: Users must set up and fund accounts with OpenAI and/or Anthropic, as well as a Stable Diffusion service, adding setup friction.
- Fantasy Genre Focus: The system is optimized for fantasy fiction; other genres may require significant prompt engineering and code modifications.
- Variable Output Quality: AI-generated prose can be inconsistent in style and depth, often requiring human editing before publication-ready results.
Frequently Asked Questions
GPT Author supports two main configurations: the original notebooks use OpenAI's GPT-4 paired with Stable Diffusion for cover art, while the newer Claude_Author.ipynb notebook uses Anthropic's Claude 3 model for improved prose quality.
A 15-chapter novel typically costs as little as $4 in API fees, depending on the model used and the length of each chapter. Costs come from your OpenAI and/or Anthropic API usage.
The tool outputs an EPUB file, which is compatible with most e-readers including Kindle (via conversion), Apple Books, Kobo, and any EPUB-compatible app.
Open the desired .ipynb notebook in Google Colab or a local Jupyter environment, add your API keys, provide your story prompt and chapter count, then run all cells. No complex local setup is required.
The codebase itself is free and MIT-licensed. However, running it requires paid API access to OpenAI and/or Anthropic, plus a Stable Diffusion image generation service.
