Frequently asked questions

Newton is an Agentic Learning Environment: a native desktop AI tutor that solves problems with real computation — symbolic math, real chemistry, statistics, your own code — instead of an LLM pattern-matching its way to something that sounds right. It also remembers your semester (documents, past mistakes, spaced-repetition flashcards) rather than starting from zero every conversation.

Yes — there's a real free tier, not a time-limited trial. There's no daily, weekly, or time-based cap on chatting with Newton on the free plan. The real difference is generation size: a single free-plan request to build flashcards, a practice exam, or a study plan aims for about 5 items, versus about 15 on Pro (the app tells you this honestly before you generate, not after). Pro adds specific extra tools on top — see the Pro section on the homepage for exactly what's gated and what isn't. Right now, Pro itself is free for students with a verified eligible school email (currently NYIT, nyit.edu) — there's no paid Pro plan for everyone yet; see the Roadmap.

It means Newton distinguishes an answer it actually computed from a judgment call it reasoned its way to, and tells you which one you got. Ask it to factor a polynomial and it runs real symbolic algebra (SymPy) to get there — the answer is derived, not recalled. Ask it to balance a chemical equation and it solves real linear algebra over the element-composition matrix (and says so honestly if the system is underdetermined instead of guessing coefficients). Ask it to check your own code and it actually runs your code against real tests in a sandbox and reports exactly which ones passed — it never just reads your code and guesses whether it looks right. When there's genuinely no algorithm for deciding something — whether a proof's logical structure holds, whether an essay argument is persuasive — Newton says so plainly instead of dressing up a guess as a computed fact.

Windows, macOS, and Linux as a native desktop app built on Tauri, plus a real browser-based version at newton.best/chat for signed-in users — no download required, and genuinely useful on a school-managed Chromebook that can't run the desktop app at all.

No. Newton's tutor, and every one of its computation tools, runs against a cloud backend — the desktop app itself doesn't do the symbolic math, chemistry, or code-checking locally. A fully offline, locally-bundled model has been discussed as a future direction but isn't built, and there's no target date for it.

Your documents and conversations are stored securely on Newton's servers so features like your persistent document library and spaced-repetition review actually work across sessions — Newton can't remember your semester without keeping that data somewhere. We don't yet have a finished, lawyer-reviewed privacy policy published on this site to link you to here; treat that as a real gap we're working through, not something we're glossing over.

Not by default, and if you turn on Focus Mode, the backend itself enforces that — not just a politely-worded instruction to the model. With Focus Mode on, Newton's server inspects its own draft reply before you ever see it and holds back a clean final answer unless you've made a real attempt or explicitly ask to be checked. get_math_hint also gives you a real three-level hint ladder (a conceptual nudge, then a first concrete step, then the answer) instead of one flat reveal.

It's a real, server-computed signal, not the model's own self-assessment. Newton's backend checks whether a reply's tool result actually came from real computation (symbolic math, chemistry's linear algebra, a sandboxed code run, a verified proof sub-step) before it ever renders the badge — a plain web search or a reasoning-based judgment call never gets one, regardless of how confident the reply sounds.

No. Flashcards, practice exams, and study plans can all be generated from just a topic — SAT prep, a subject you're learning from scratch, anything — with no document required. Uploading your own syllabus, readings, or notes makes everything more specific to your actual coursework, but it's not a hard requirement to get started.