GitHub Spark allows developers to build and deploy full-stack, LLM-powered apps on the GitHub platform by describing their idea for an app.
GitHub Spark, an AI-powered tool for building full-stack intelligent apps, with front-end, back-end, and AI features included, using only natural language, has moved to a public preview stage.
The public preview, currently open only to Copilot Pro+ users, was announced July 23. Plans call for extending the preview to additional customers soon.
With GitHub Spark, developers describe their idea for an app and Spark builds it, front-end and back-end capabilities included, GitHub said. No setup or configuration is required, according to GitHub. Developers can visit github.com/spark to build their first app. Access to Spark also can be enabled by signing up for a Copilot Pro+ account.
GitHub Spark had been in a technical preview since October 29, 2024. Key features listed for Spark include the following:
- Natural language app creation, powered by the Claude Sonnet 4 hybrid reasoning model.
- Data storage, LLM (large language model) inference, hosting, deployments, and GitHub auth included out-of-the-box.
- Add intelligent features powered by LLMs from OpenAI, Meta, DeepSeek, xAI, and other providers, with no API key management needed.
- One-click app deployment.
- Use natural language, visual editing controls, or code with GitHub Copilot code completions to iterate on ideas.
- Create a repository with GitHub Actions and Dependabot.
- Open a codespace directly from Spark to iterate with Copilot agent mode or assign an issue to Copilot coding agent.
The public preview of GitHub Spark follows the introduction last week of Google Labs’ Opal, which is a tool for building AI-powered apps using natural language commands or a visual editor, or both. Opal is available in a public beta in the United States.
