Heptabase Review, Pricing & Alternatives (2026)

Heptabase is made by Hepta Platforms, Inc. and is an actively developed visual knowledge-management service. Its focus is research and learning rather than general project management. The product combines reusable cards, PDFs, journals, databases, tags, research imports, AI chat, and agent workflows. Desktop storage has local-first and offline behavior, while optional cloud sync works across unlimited devices. Mobile supports capture and reading. Heptabase is proprietary software, not an open-source project, and its hosted design means cloud processing remains part of the service even when one device holds a local database. The company describes its early-bird offer as ending on August 31, 2023, with version 1.0 launching at that time.

A typical Heptabase workflow starts with PDFs, web or video content, notes, and highlights. Documents can be parsed or processed with OCR, after which selected passages become reusable cards. Users arrange cards on nested whiteboards, connect them with tags, and search the resulting knowledge base. Cards can appear on multiple boards, which suits projects that revisit the same source material. AI Tutor, chat, transcripts, Insight Generator, and voice-note features add assistance around collected material. Users can publish or share selected spaces. PDF extraction creates text for AI, including scanned text through OCR, but it does not automatically create highlight cards. Users must select passages when they want highlight cards.

Heptabase added source question-answering modeled after NotebookLM in December 2025, placing AI across notes, whiteboards, PDFs, and video workflows. The same release described self-hosted parsing and embeddings, meaning Heptabase runs those internal models rather than offering an installable server. Its 2025 changelog lists GPT-5, GPT-5 Nano, GPT-5 Mini, Gemini 2.5 variants, and Claude 4.1 Opus, with model rotation. December added Gemini 3 Flash and GPT-5.2. The 2026 changelog lists GPT-6 Sol, GPT-6 Luna, Claude Opus 5.5, and Gemini 4 Argon, plus expanded MCP access to original attachments. Model availability can change, and new provider models are not automatically included.

The seven-day trial includes unlimited PDF uploads and parsing, plus the Premium AI credit allowance for up to seven days or until trial credits run out. Pro costs $11.99/month or $107.88/year, with 100 AI credits monthly and high PDF limits. Premium costs $23.99/month or $215.88/year and includes 1,800 monthly credits, unlimited PDF uploads and parsing, and broader model access. Premium+ costs $71.99/month or $647.88/year with 8,100 monthly credits. Each PDF is limited to 350MB and 3,000 pages. Parsing does not consume credits. Credits cover agents, Tutor, transcripts, insights, and voice notes. Paid users can enable separate on-demand billing.

Heptabase's main tradeoff is cost and service dependence. Premium and Premium+ meter AI work through monthly credits, there is no end-to-end encryption or supported self-hosted Heptabase server, and server-side indexes contain content chunks, metadata, and vectors. Obsidian offers local Markdown and optional Sync, while Heptabase provides a stronger built-in PDF, card, and whiteboard workflow. Scrintal is stronger for real-time collaboration, templates, and tablet ease, while Heptabase is more research-focused. Notion is better for team databases and project management. Logseq offers an open-source, outliner-oriented alternative with stronger local ownership. Pick Heptabase if spatial source synthesis matters more than local ownership or low subscription cost.