A calm writing studio where AI helps you hold an entire story together — from the first idea down to every page. Local or cloud. Your plot, your control.
Free with local models · No account needed to try
The harbour had gone quiet by the time she reached the last pier, and the lamps were only beginning to find their reflections in the water.
Ask a model for a novel and you get beautiful paragraphs that forget what happened three chapters ago. PenFreely solves the problem of scale — it plans the whole book first, then writes prose page by page, always aware of what came before.
Book, part, chapter, section, page. Each level inherits intent from the one above, so every page knows its purpose. Prose is generated only at the page level — grounded in the plan and brief summaries of neighbouring pages.
The cascade keeps characters, threads and tone consistent across hundreds of pages — not just within a paragraph.
Run local models with Ollama or reach for the cloud via OpenRouter. Switch providers — or assign different models to planning, prose and summaries — at any moment.
With local models your manuscript never leaves your machine. No training on your words, no account required to begin.
PenFreely shows live cost, tokens per second and which models did what — broken down by chapter and page. Cloud models are capped at $0.20 per million tokens unless you say otherwise.
It feels less like prompting a machine and more like keeping a long, careful conversation with my own book.From an early reader
Bring an idea. Leave with a structure, a cast, and your first chapter.
Start writing