Bolt.diy

AI Coding ToolsAI Code EditorsVibe Coding Tools
  1. About
  2. Pricing
  3. What people are saying

About Bolt.diy

Bolt.diy is a powerful open-source community fork of Bolt.new, designed for developers seeking a flexible, self-hosted AI development environment. Unlike proprietary alternatives like v0 or Lovable, Bolt.diy is MIT-licensed and allows users to bring their own API keys from 19+ providers (now integrated via the DaoXE multi-protocol gateway), or run models locally via Ollama.

This approach provides unparalleled control over costs and data privacy. Featuring tools like file locking, MCP support, long-term memory for larger projects, and integrations with Supabase and GitHub, Bolt.diy synthesizes local development comfort with cloud power to simplify building full-stack applications.

Pricing

  • Free - Open-source software distributed under the MIT license.
  • Self-hosted - No subscription fees; users maintain complete control over their deployment environment.
  • Bring Your Own Keys - Users pay only for LLM usage via their own API providers.
  • Local Models - Supports free local inference via Ollama or LM Studio for zero-cost generation.
  • Commercial Licensing - Requires a WebContainer API license for production use in for-profit settings.

What people are saying

Users praise Bolt.diy for its flexible self-hosted model, but note it requires more technical expertise to deploy and sometimes struggles with large codebases.

  • "Bolt.diy is fussier, harder to deploy and doesn’t come together nearly as fast. I think you would need to have more technical experience to use diy [than the official version]…" — @isigg9 (September 2026)
  • "if you have a BIG project dont even bother with bolt.diy its completely unusable… ive tried different models and it somewhat worked with Gemini 2.0." — @ilfzji (September 2026)
  • "We don't have a commercial offering yet and are planning to migrate off WebContainers for the upcoming full stack features -- WebContainers show their limits pretty quickly…" — @bhackett (September 2026)