Presenton: The Free, Self-Hosted Gamma Alternative With an API
Presenton — a self-hostable AI presentation generator. Give it a prompt, a document, or your own outline, and it builds a full slide deck, exportable as editable PowerPoint or PDF, with a REST API and an MCP server for automated generation.
Alternative to: Gamma Plus, $9/seat/month ($108/yr billed annually), 1,000 credits/month, up to 100 slides per prompt. Gamma Pro, $18/seat/month ($216/yr), 4,000 credits/month, API access. Beautiful AI Pro, $14.50/mo billed annually. Decktopus Pro, $9.99/mo billed annually ($119.88/yr).
Cost to run: free software, bring-your-own-key. You pay only for the LLM and image-API calls it makes — or $0 if you run local models through Ollama or LM Studio.
Licence: Apache-2.0, confirmed by reading the full text. No commercial-use restrictions. "Presenton Cloud" and "Presenton Enterprise" are separate proprietary offerings sitting alongside the open-source core.
Runs on: Docker (docker run ghcr.io/presenton/presenton:latest), or as an Electron desktop app for Mac, Windows, and Linux.
Requires: an API key from one of over a dozen supported LLM providers — OpenAI, Anthropic, Gemini, Azure, Bedrock, OpenRouter, and others — or none at all if you run Ollama locally.
Traction: 10,290 stars · 1,584 forks · 56 watchers · 194 commits in the last 30 days, 1,056 in the last 90.
Liveness: very live — last commit 3 days ago, near-200-commits-a-month pace, run by a core team of three (Saurav Niraula, Shiva Raj Badu, Sudip Parajuli) plus outside contributors.
Maturity: v0.9.10-beta — still tagged beta despite the traction and commit volume. 83 tags.
Links: repo · site
Figures as of Sept. 17, 2026.
What it does
Presenton takes a prompt, an uploaded document, or a slides_markdown outline you write yourself, and turns it into a full slide deck. It ships six built-in templates and lets you turn your own PowerPoint design into a reusable AI template. Output comes back as an editable PowerPoint file or a PDF, and there's a browser-based drag-and-edit interface if you want to tweak the result by hand.
It's built to be automated, not just clicked through: a REST API (POST /api/v1/ppt/presentation/generate) lets you generate decks programmatically with parameters for tone, slide count, and even live web-search grounding, and a built-in MCP server means Claude or another MCP-compatible agent can generate a deck directly. It also ships as a standalone Electron desktop app for Mac, Windows, and Linux — though the MCP server is turned off in that build to avoid authentication conflicts.
What it replaces, and what that costs
Presenton names its own competition in its README title: it calls itself an alternative to Gamma, Canva, Beautiful AI, Decktopus, and Presentations AI. Checked directly on the vendor pricing pages: Gamma's Plus tier is $9 a seat per month ($108 a year billed annually) for 1,000 credits a month and up to 100 slides per prompt; its Pro tier, marked "most popular," is $18 a seat per month ($216 a year) with 4,000 credits and API access; its top Ultra tier runs $90 a seat per month. Beautiful AI's individual Pro plan is $14.50 a month billed annually, and its Teams plan is $40 a month per user. Decktopus's cheapest annual plan works out to $9.99 a month ($119.88 a year) for 750 AI credits, about 150 slides.
Presenton itself is Apache-2.0, self-hosted, and bring-your-own-key — its own pitch is "only pay for what you use, no hidden fees or subscriptions." The contrast isn't a specific dollar figure; it's the difference between a recurring per-seat subscription that runs forever and paying only for the tokens you actually consume, which can be $0 if you run a local model through Ollama.
Is it still alive?
Very. Presenton is shipping close to 200 commits a month, with the most recent commit three days before this was written. It's run by a small core team of three — Saurav Niraula, Shiva Raj Badu, and Sudip Parajuli — who account for the large majority of its roughly 2,780 total commits, with community contributors filling in the rest. That's a real team, not a single maintainer's side project.
One thing worth noting: the latest release is still tagged v0.9.10-beta. For a project with 10,000-plus stars and daily commits, staying in beta versioning this long is either genuine caution or a sign the maintainers consider it not fully settled yet — worth knowing if you're deciding whether to build something on top of it.
What people say
Hacker News discussion is minimal — two low-traffic Show HN posts, "Built an Open Source BYOK Gamma Alternative for AI Presentations" (2 points) and "Open-Source API for AI Presentation Generation" (2 points, 2 comments, with the founder replying about an early pivot away from Electron toward developer tooling). For a repo with this much GitHub traction, that's thin: whatever conversation exists is mostly happening on LinkedIn and YouTube, not HN. A small YouTube review (under 2,000 subscribers, under 1,000 views) covered it as an open-source alternative worth trying, but there's no meaningful independent community discussion to quote.
No verified official X or Twitter account exists for the project either — it's reachable through its Discord, YouTube channel, and company site, but not through a verified social handle.
The catch
The slides_markdown input path is explicitly not deterministic — the README says plainly that it "is not a deterministic Markdown-to-slide renderer," and your text may get rephrased to fit slide templates, with a 100-word-per-slide and 50-slide cap by default. Presenton also publishes no cost estimator or budget guardrail for the LLM calls it makes on your behalf — you're fully exposed to whatever your chosen provider charges, with no dashboard to track it.
The MCP server — one of the two features that make this different from a consumer deck tool — is disabled in the desktop app and only works in the self-hosted Docker deployment.
Credits
Built by Saurav Niraula, Shiva Raj Badu, and Sudip Parajuli, with community contributors. Figures pulled from ungh.cc, a full clone of the repository, and the project's own README and LICENSE. Pricing verified directly on Gamma's, Beautiful AI's, and Decktopus's own pricing pages.