Mac AI AssistantUpdated August 2026

Elephas Review 2026: Pricing, Features, Pros & Cons

Item: Elephas
Rating: 4.3
Author: AISO Tools

Elephas is built for the work you cannot paste into ChatGPT. Everything about it — local indexing, redaction of 28 sensitive entity types, a fully offline mode — follows from that one constraint. Which makes it excellent for a narrow audience and overpriced for everyone else. Here is the honest read on the three tiers, what Super Brain actually gets you, and when a free alternative is the right answer.

Quick Verdict

4.3/5

Overall Rating

28 types

Redacted before cloud calls

$39/mo

Real entry price (Super Brain)

Best for: consultants, lawyers, accountants and researchers working under NDA on a Mac. Not for: anyone without a confidentiality constraint — you are paying a privacy premium for a problem you do not have.

What Is Elephas?

Elephas is a Mac-native AI assistant whose entire architecture is organised around one problem: a large number of professionals cannot legally or contractually paste their working material into a mainstream chatbot. Client files under NDA, privileged legal documents, financial records, unpublished research. For those people the question is not which assistant is smartest — it is which assistant they are permitted to use at all.

The answer Elephas gives is a stack of controls rather than a blanket refusal to touch the cloud. Your documents, notes and files are indexed 100% locally on the machine. An offline mode runs fully on-device. When you do choose a cloud model, the app redacts 28 categories of sensitive data before anything leaves the Mac, and the calls that do go out hit zero-data-retention endpoints. Model choice stays yours — GPT-5, Claude and Gemini are all available — so you are not trading capability for compliance.

Functionally it does AI chat and writing across the whole Mac, works against your PDFs, notes and files, and offers document and note workflows with an iOS companion. The feature that distinguishes it in daily use is Super Brain: a persistent knowledge base assembled from your own material, queried with citations back to the source documents. It requires macOS 13 or later, and every new download activates seven days of Pro automatically with no credit card.

Elephas Pros & Cons

✓ Pros

•The privacy architecture is a set of controls rather than a marketing claim: 100% local indexing of your files, automatic redaction of 28 sensitive entity types before any cloud call, zero-data-retention endpoints when cloud models are used, and a fully on-device offline mode
•You are not locked to one model — GPT-5, Claude and Gemini are all selectable, so the privacy posture does not cost you frontier capability the way a purely local-only tool would
•Super Brain is the feature people actually stay for: a persistent knowledge base built from your own documents, notes and files, with answers cited back to the source, rather than re-uploading context every session
•It works across the whole Mac rather than inside one app window, so the assistant is available where the writing is happening instead of in a separate tab
•Integrations with Apple Notes, Notion and Obsidian mean it reads the notes system you already have instead of asking you to migrate into a new one
•Every new download activates 7 days of Pro automatically with no credit card, which is a genuinely low-friction way to find out whether the Super Brain workflow fits how you think
•Super Brains are accessible on iPhone and iPad from Professional up, so the knowledge base is not stranded on the desktop
•Pro+ ships a full redaction inspector showing the exact entities stripped per message — for anyone who has to defend a tooling choice to a compliance officer, that audit surface is the whole argument

✗ Cons

•macOS 13 or later only. There is no Windows or Linux version, and the iOS apps are companions to a Mac install rather than a standalone product
•There is no free tier — only a 7-day Pro trial. After that the entry price is $19/month, which is above the $0 that a great many people are anchored to for an AI chat interface
•$49/month for Pro+ is steep against general-purpose assistants, and the pricing page's own justification (it pays for itself in 15 minutes of a billable hour) tells you exactly who it is priced for — if you do not bill hourly, that logic does not transfer
•The redaction visibility that makes the privacy story defensible is tier-gated: basic visibility on Professional, the full inspector and advanced redaction configuration only on Pro+
•Super Brain — the differentiating feature — is not in the $19 Standard tier at all, so the entry plan is a fairly ordinary AI writing assistant with good privacy defaults
•Offline AI workflows are Pro+ only, which is awkward given that offline capability is one of the headline reasons to choose the product
•The “maximum” usage on Pro+ is governed by a fair-usage policy granting 1,000 credits per month — a soft ceiling rather than the unlimited the tier name implies
•Checkout runs through Gumroad rather than a first-party billing portal, which is fine functionally but reads as less enterprise-ready than the Pro+ positioning suggests

Elephas Pricing 2026

Entry

Standard

$19/mo

•AI chat and writing across your Mac
•Basic document and note workflows
•Works with your PDFs, notes and files
•Cloud AI with zero data retention
•Local-first indexing, automatic redaction

Knowledge workers who want a private assistant, not a knowledge base

Professional

$39/mo

•Everything in Standard
•Super Brain — chat with your own knowledge
•Apple Notes, Notion, Obsidian integrations
•Super Brains on iPhone and iPad
•Basic redaction activity visibility

Daily work across a real corpus of documents and notes

Best for Sensitive Work

Pro+

$49/mo

•Everything in Professional
•Full redaction inspector, per-message detail
•Advanced redaction configuration
•Optional offline AI workflows
•1-on-1 onboarding, priority support

Client work under NDA where you must evidence what was redacted

Annual billing saves roughly 17%. No permanent free tier — every new download includes 7 days of Pro, activated automatically, no credit card. Pro+ "maximum" usage runs under a fair-usage policy granting 1,000 credits per month. Requires macOS 13+. Plan contents as published on Elephas's pricing page in August 2026 — verify before purchase.

Who Should Actually Use Elephas

Use it if: you work on a Mac with material you are contractually or professionally barred from putting into a mainstream assistant; you need a persistent knowledge base over your own documents rather than per-session uploads; or you have to be able to show someone — a compliance officer, a client, a regulator — exactly what was redacted before a request left your device, which is the Pro+ redaction inspector and nothing else on this list does it as legibly.

Skip it if: your work is not confidential, in which case a general assistant is cheaper and more capable; if you are not on macOS 13+, since there is no other platform; if $39/month is the ceiling and you only need the Standard feature set, because Standard without Super Brain is a fairly ordinary writing assistant; or if your requirement is that nothing at all leaves the machine, in which case price the Pro+ tier from the start rather than discovering that offline mode is gated.

Frequently Asked Questions

What does Elephas actually do that ChatGPT does not?

Three things, all downstream of one constraint. First, it indexes your documents, notes and files entirely on your Mac, so the corpus is never uploaded — you get a persistent knowledge base instead of re-pasting context into every session. Second, it redacts 28 categories of sensitive entity (names, emails, phone numbers, addresses, financial identifiers and so on) before any request leaves the machine, and the cloud endpoints it does call are zero-data-retention. Third, it has a fully on-device offline mode for material that should not touch a network at all. If none of those constraints apply to your work, ChatGPT is cheaper and more capable and you should use it.

How much does Elephas cost in 2026?

Three tiers, monthly: Standard at $19, Professional at $39 and Pro+ at $49, with roughly 17% off for annual billing. There is no permanently free plan — every new download activates seven days of Pro automatically, no credit card, and then you pick a tier. The important detail is that the tiers are not just usage ladders: Super Brain, the persistent knowledge base that is the actual reason to buy this product, starts at Professional, and offline AI plus the full redaction inspector are Pro+ only. Budget from the feature you need rather than from the headline price.

What is Super Brain?

A persistent, queryable knowledge base built from your own material — PDFs, notes, files, and connected sources like Apple Notes, Notion and Obsidian. Rather than uploading documents per conversation, you build a Brain once and then ask it questions, with answers cited back to the specific documents they came from. It is available from the Professional tier upward and accessible from iPhone and iPad as well as the Mac. In practice this is the feature that separates Elephas from a well-configured ChatGPT workflow, and it is why the $19 tier feels thin by comparison.

Is Elephas genuinely private, or is that marketing?

The claims are specific enough to check, which is a good sign. Indexing happens locally. Redaction covers a stated 28 entity types and runs before the request leaves the device. Cloud calls go to zero-data-retention endpoints. Offline mode runs fully on-device. What you should be clear-eyed about is that using a cloud model at all means a redacted version of your text reaches a third party — the redaction is a risk reduction, not an elimination. If your obligation is that no derivative of the material may leave the machine, the only compliant configuration is Pro+ with offline workflows, and you should price accordingly.

Who is Elephas actually built for?

Professionals whose material is under NDA or professional privilege and who therefore cannot use mainstream assistants at all: consultants, lawyers, accountants, financial advisors and researchers. The pricing page makes this explicit — Pro+ is framed as paying for itself in under 15 minutes of a single billable hour. That framing is the tell. If you bill for your time and your documents are confidential, the maths is easy. If you are a student, hobbyist or a developer who mostly works on public code, you are paying a privacy premium for a constraint you do not have.

What are the best Elephas alternatives?

For local-first knowledge bases on a Mac, Obsidian with a local LLM plugin gets you most of the Super Brain concept for free if you are willing to assemble it. NotebookLM is stronger at document synthesis and citation but is cloud-only and Google-hosted, which is disqualifying for exactly the audience Elephas targets. Msty and LM Studio give you a polished interface over fully local models with no cloud path at all — more private, less capable. And if the requirement is really just a Mac-native assistant with a good keyboard flow rather than a privacy mandate, Raycast AI is cheaper and faster to live with.