Presenton is an open-source, self-hostable AI presentation generator that positions itself as a direct alternative to Gamma, Beautiful AI, and Decktopus. With 6,419 GitHub stars and 1,100+ forks, the project surged onto GitHub Trending this week as developers and small teams looked for a way to escape per-seat SaaS pricing and bring-your-own-key flexibility for slide generation. The repository ships under Apache 2.0 with a web app, an Electron desktop app for Windows, macOS, and Linux, a REST API, and Docker images, covering essentially every deployment shape that an enterprise might want. ## Bring-Your-Own-Key for Every Major LLM The key architectural choice that separates Presenton from incumbent SaaS tools is provider neutrality. The generator integrates with OpenAI, Google Gemini, Vertex AI, Azure OpenAI, Anthropic Claude, Amazon Bedrock, Fireworks, Together AI, LM Studio, Ollama, and any OpenAI-compatible endpoint. That means a team can run privacy-sensitive presentations entirely against a local Ollama or LM Studio model, use Claude Opus for the highest-quality decks, or route through Bedrock for compliance reasons, all without changing how the generator is wired into the rest of the workflow. ## Structured Generation Controls The generation interface exposes the dials that most AI slide tools hide. Slide count is configurable with a default of 8, tone options include default, casual, professional, funny, educational, and sales pitch, and verbosity can be set to concise, standard, or text-heavy. Users can supply custom slide markdown as an alternative to letting the model auto-outline, and an optional web-search grounding step pulls in fresh facts so a quarterly business review or product launch deck reflects current information rather than the model's training cutoff. ## Editable PPTX and PDF Output The output story is where Presenton differentiates from many open alternatives that only export to images or HTML. Decks render to fully editable PowerPoint (PPTX) files plus PDF, so the generated deck can be polished in Keynote or PowerPoint instead of being a write-only artifact. Templates and themes are authored in HTML and Tailwind CSS, which means designers can build a brand-specific template once and have every generated deck inherit the corporate look. The project also supports template generation from an existing PowerPoint document, letting a team reverse-engineer their preferred layout into a reusable Presenton theme. ## Deployment Across the Stack The desktop application uses Electron and runs natively on Windows, macOS, and Linux with no browser required, ideal for sales teams that want a single icon on their laptops. Docker images with GPU support cover self-hosted server deployments, and the repository documents one-click Railway and DigitalOcean deployment paths for teams that want a managed cloud experience without giving up source-code access. A REST API on top of all of this turns Presenton into a back-end service that other applications can call to generate decks programmatically, useful for sales enablement platforms, learning management systems, and reporting tools. ## Honest Tradeoffs Presenton is genuinely usable in production today, but it is not a drop-in replacement for Gamma's polish. The default templates are functional rather than beautiful and require some HTML and Tailwind work to reach brand-grade quality. Image generation and richer chart types are evolving, with the current release leaning on stock icons and Tailwind-styled graphics rather than full AI-generated imagery. For teams that value data residency, model choice, and zero per-seat fees more than out-of-the-box visual polish, Presenton is now the leading open option in a category that until very recently had no credible self-hosted alternative.