Nomi Review: Memory, Relationships & Pricing

Nomi is an AI companion platform for romantic, friend, mentor, and custom relationships rather than only an AI-girlfriend use case. It documents memory systems, group chats, images, video, voice, calls, internet access, and proactive messages across web and mobile channels.

This Research Review analyzes the prepared first-party record. CompanionScope has not yet performed a hands-on Scope Test of Nomi’s conversation, memory, or media behavior.

What is Nomi?

Nomi’s relationship framing is comparatively broad. Users can create romantic, friendship, mentor, or custom companion dynamics, supported by conversation, multiple Nomis, group chats, and media features.

Web, iOS, Android, and PWA access are documented as available. That broad channel coverage is a documented fact; consistency among those channels remains untested.

Subscription pricing

The current first-party record lists $15.99 monthly, $39.99 quarterly, and $99.99 annually. Free access is documented without requiring a payment card, although exact supplied limits across all capabilities are incomplete.

Paid accounts can have up to 10 active Nomis and 10 active group chats before purchasing additional slots. An active paid subscription is required before credits can be purchased. Those dependencies matter because subscription price is only the base layer for people who use extra slots or credit-metered features.

Credits and incomplete True Cost

Nomi documents one credit for a selfie request, art request, or image-edit request. Additional Nomi slots cost 100 credits, additional group-chat rooms cost 25, standard-definition video requests cost 6, and high-definition video requests cost 10.

Current USD prices for credit packs were not sufficiently verified. The usage schedule therefore cannot be translated into a reliable dollar-based monthly total. The subscription prices are documented, but True Cost remains Partial for credit-dependent use.

Memory and Mind Maps are strong claims

Nomi documents short-, medium-, and long-term memory, Mind Maps, and persistent relationship continuity. It also uses the marketing phrase “human-level memory.” CompanionScope treats all of these as first-party claims rather than performance results.

A future Scope Test should examine factual recall, correction handling, contradictions, continuity after time away, and what Mind Maps expose or let users control. It should also distinguish memory that feels coherent from memory that is merely repeated in marketing language.

Conversation and relationship breadth

Conversation, proactive messages, internet access, group chats, and multiple relationship modes are documented. Their presence may support different use cases, but the current research cannot establish tone quality, initiative, emotional consistency, or the usefulness of mentor and friend configurations.

Group-chat limits and extra-room costs are documented commercial facts. Context-sharing behavior among multiple Nomis remains a hands-on question.

Images, video, voice and calls

Images, image editing, video, voice, and voice calls are documented. Credit use is published for several image and video actions, but output quality, retry rates, generation time, voice naturalness, and call stability have not been measured.

The difference between standard- and high-definition video pricing is visible in credits. Whether the quality difference justifies the extra consumption is outside this Research Review.

Adult text and explicit images differ

Nomi documents adult or unfiltered text conversation and roleplay as available. It separately states that nude or NSFW image generation is unavailable. These are distinct capability decisions and should not be collapsed into a single NSFW label.

The Research Snapshot and capability details preserve that distinction: documented adult text does not imply documented explicit-image generation.

Privacy is more nuanced than one claim

First-party material describes conversations as private or anonymized in relevant contexts, while message data is processed to generate replies and some data can be used to improve Nomi. The current material does not establish end-to-end encryption; it references de-identification workflows and training archives.

Account deletion generally removes personal data, subject to documented legal, operational, or retention exceptions. These policies do not support an absolute claim that no data is retained or used, and they do not establish how easy the live controls are to use.

What remains unknown

Open questions include:

• a reconstructed universal current checkout across channels;
• current USD credit-pack prices;
• complete free limits across conversation and media;
• actual memory and Mind Map performance;
• conversation quality across relationship modes;
• proactive-message, group-chat, voice, call, and video behavior;
• adult-text boundaries in practice; and
• live privacy, deletion, and data controls.

Research Review conclusion

Nomi combines broad relationship positioning, wide channel access, published subscription prices, and unusually detailed memory claims. Its adult-text and image rules are also clearer when treated as separate capabilities.

The major unresolved questions are experiential and economic: memory performance has not been tested, and documented credit use cannot be converted into dollars without verified pack prices. No Scope Score or recommendation follows from the document record alone.