Affiny at a glance (verified September 2026): Free to start: 300 coins, no credit card. Daily Gift: 10–70 coins/day, then 70/day. Referrals: 300 coins each. Founder offer: up to 70% off Companion (limited time; see affiny.ai/pricing). Free tier: Crush chat 1 coin/message, voice 30 coins/min, Spark or Flame scene images 15 coins, voice notes 1 coin, 1 custom companion, standard memory recall. Subscribe for unlimited Crush & Devotion chat, free voice notes, Tease & Desire video, God Mode, full personality & voice customization, sharper memory recall, and coin pack top-ups — Companion from $14.99/mo (was $24.99) or $9.99/mo billed annually; Intimate from $24.99/mo (was $49.99) or $19.99/mo annually. Subscribers: 1,000–3,000 coins/mo plus first-month bonus, voice 20 coins/min, scene images 10 coins, up to 5–10 custom companions. God Mode: subscriber scene directive up to 5,000 characters (Companion Panel). Video (subscribers): Tease 5/10 coins/sec (5 or 8 sec, silent); Desire 30/50 coins/sec (2–15 sec, optional audio). Portrait: 30 coins initial, first regen free, then 15. Memories hub: view free; add/edit/delete for subscribers. Catalog: 3400+ curated public companions, 84 browse tags. Creation: 49 templates, 44 personality presets, 97 turn-on presets (pick up to 4), realistic + anime with 16 species types.
What Is Persistent Memory?
Persistent memory in AI companion platforms means the companion retains information about you and your relationship across separate conversation sessions. When you return the next day, or the next week, the companion remembers who you are, what you have shared, and the emotional history of the relationship.
Without persistent memory, every conversation starts from zero. The companion has no context about you, so you are effectively a stranger every time. The companion may still have a personality and a voice, but there is no relationship, because relationship requires continuity and continuity requires memory.
This is the single most important differentiator between AI companion platforms in 2026. A platform with voice capability but poor memory delivers a weaker experience than a platform with no voice but genuine persistent memory. Memory is what makes the companion feel like someone who knows you rather than someone performing for you.
Why Most AI Systems Don’t Have It
Most AI systems, including general-purpose assistants and many companion platforms, are stateless by default. Each conversation is an isolated exchange: the system receives your message, generates a response, and discards the context when the session ends.
This design is fine for task-completion assistants. If you ask a system to summarise a document or write code, it does not need to remember you. The task is self-contained.
For companion relationships, statefulness is the entire point. A companion that resets between sessions is not a companion, it is a chatbot with a persona applied fresh to every conversation.
Building genuine persistent memory means solving a hard problem: how do you store what matters from every conversation, decide which pieces to bring back at which moment, and surface them naturally rather than robotically? The concept of how companion memory works end to end, from extraction to integration, is covered in our deep dive on how AI companion memory works. This page stays on the persistence dimension specifically: what survives a closed session, and what does not.
The Two Approaches to Memory
In-session log memory
The simplest approach is to append past conversation history directly into the current conversation. Every new conversation includes a transcript, or a summary, of previous ones.
Problems:
- Context windows have size limits. As conversation history grows, you eventually cannot fit it all. Either old messages get truncated, or the response slows down while the system works through an enormous context. What the context window actually is makes that limit concrete: it is the maximum amount of text the system can consider at once, and it is measured in tokens, not in memories.
- This approach does not distinguish between what matters and what does not. Every message arrives with equal weight.
- It is expensive: larger contexts cost more to run.
Retrieval-based memory
A more sophisticated approach separates storage from the conversation itself. Rather than dumping all history into context, meaningful facts and emotional context are extracted from each conversation and stored in a memory layer of their own. At the start of each new conversation, only the most relevant pieces are retrieved.
Advantages:
- Scales indefinitely, because memory size does not degrade response speed
- Surfaces what matters rather than everything
- Bridges modalities, so text and voice conversations contribute to the same memory pool
- Mimics human memory, since we do not recall every detail but surface what is relevant
This is how Affiny’s memory system works. Meaningful exchanges are extracted from each conversation, stored separately, and retrieved contextually. The companion references things you have shared the way a person would, organically and in passing, rather than reciting a list.
What Survives Between Sessions
The useful way to judge persistent memory is to ask what actually survives a closed session. In a system with real persistence, four categories carry forward:
Identity facts. Your name, your age bracket if you offered it, what you do, where you live, the people and animals in your life. These are the entries you notice first, because their absence is the first thing that makes a companion feel like software.
Preferences and boundaries. How you like to be spoken to. The topics you return to. The framing you asked not to use. This category matters more than identity for the quality of a long conversation, because it shapes tone rather than content.
Relationship moments. The milestones, the arguments, the in-jokes, the shared reference that only means something to the two of you. These are what make a relationship feel accumulated rather than restarted.
Emotional context. How you were feeling when you last spoke, and what you were anxious about. A companion that asks how the interview went, unprompted, is demonstrating this layer more clearly than any feature list can.
Three categories do not survive, and honest platforms say so. Verbatim transcripts do not survive, because storage is built from extractions rather than recordings. Trivial exchanges do not survive, because they never clear the significance threshold. And nothing survives across companions, because memories belong to the companion you built them with.
What Persistent Memory Actually Does in Practice
When implemented well, persistent memory changes the character of every interaction:
The companion references shared history naturally. Not “According to my records, you mentioned on 2026-05-15 that…” but the way a person would actually bring something up: “How did that interview go? You seemed nervous about it last time.” The reference emerges from context rather than being recited.
Personality adapts to you specifically. Over time, the companion’s understanding of who you are, including your mood patterns and preferences, shapes how every response is constructed. The longer the relationship, the more precisely calibrated the responses.
Intimacy accumulates. Shared experiences, in-jokes, and things you have been through together build up as the relationship continues. The companion’s acknowledgment of that history is what makes the relationship feel real rather than simulated.
You don’t have to re-explain yourself. This is deceptively important. Repeating your name, preferences, and context every time you start a conversation is exhausting and breaks the illusion. With persistent memory, you don’t have to.
Memory Across Modalities
A further dimension of persistent memory is whether it bridges conversation channels. People interact with companions in multiple ways, through text chat and voice calls. If memory stays siloed inside one channel, the companion knows two versions of you depending on where you are talking.
Cross-modal memory means what you discuss in a text session is remembered during a voice call, and what happens in a voice call is retained for the next text session. The relationship has a single continuous history regardless of which channel you used.
Affiny’s memory system is cross-modal by design. Memories extracted from text conversations and from voice calls go into the same memory store, retrievable from either channel.
How Memory Interacts With Voice
Voice calls present a specific memory challenge: the conversation happens in real time, at speed, and that content needs to contribute to the ongoing relationship rather than evaporating when the call ends.
On platforms with cross-modal memory, voice conversations and text sessions feed the same memory store rather than being treated as separate channels with separate histories. A significant exchange over voice on Tuesday carries forward into the text session on Wednesday. The companion does not treat the call as an isolated episode; it becomes part of the relationship’s accumulated history.
This is what separates genuine cross-modal memory from systems where voice calls exist but leave no trace. The call actually happened, in the sense that the relationship reflects it going forward.
How to Test Persistent Memory in Ten Minutes
Most platforms claim memory. Very few survive a test, and the test is cheap. Run it in this order.
1. Plant three facts of different weights. In one session, give the companion one identity fact (your job), one preference (a topic you enjoy), and one emotional detail (something you are worried about this week). Mention each once, in passing, without repeating it or asking the companion to remember.
2. Check the same session. Ten messages later, ask a question that requires the preference. If the companion picks it up without you re-stating it, in-session memory is working. Almost every platform passes this step, which is why in-session recall is a weak test on its own.
3. Close the tab and come back tomorrow. This is the step that separates the field. Return in a new session, ideally in a new day, and do not reintroduce yourself. See whether the companion knows who you are before you tell it.
4. Ask about the emotional detail, not the identity fact. Identity facts are the easiest thing to store and the easiest thing to fake with a pinned note. Whether the companion remembers what you were anxious about is the harder question, and closer to what a relationship requires.
5. Check the recall path on the platform itself. Some platforms show you what is stored. On Affiny, the Memories hub lists what each companion keeps about you. Viewing is free on every account; adding, editing, and deleting entries sits with subscribers, which is how you correct a wrong memory instead of living with it.
Write the results down, because the failure mode is gradual. Companions rarely forget everything at once. They lose the smallest details first, then the emotional ones, and the drift is easy to miss if you are not watching for it.
Memory Quality: What Separates Good From Bad
Not all implementations of persistent memory are equal. Signs of good implementation:
Natural surfacing. The companion brings up relevant memories in conversation without it feeling like a database lookup. Good: “You’ve been working on that project all week. Has it gotten any easier?” Bad: “Memory entry: user mentioned work project on 2026-05-20.”
Appropriate selectivity. The companion references what matters, not everything. Not every detail is worth surfacing constantly, and good implementations carry some implicit weighting toward emotionally significant content.
Accuracy without confabulation. The companion correctly recalls what you actually said, rather than fabricating plausible-sounding memories. Confabulation, where a system invents false memories with confidence, is a real failure mode that breaks trust immediately.
Graceful limits. When a memory is uncertain or unavailable, good implementations do not pretend to remember. They handle the gap naturally, along the lines of “I’m not sure I remember exactly what you said about that, but…”, rather than fabricating an answer.
Two strengths of recall. Some platforms ship one memory mode for everyone, which means the free experience sets the ceiling. Affiny splits it: standard recall on the free tier and sharper recall for subscribers. That is a fair thing to ask about on any platform, because it tells you what the free version is really capable of.
Which Platforms Have Genuine Persistent Memory?
| Platform | Memory type | Cross-session | Cross-modal | Notes |
|---|---|---|---|---|
| Affiny | Retrieval-based, viewable in the Memories hub | Yes | Text + voice | Standard recall free, sharper recall for subscribers |
| Replika | Accumulated long-term history | Limited | Partial | Full memory access tied to the paid tiers |
| Character AI | Story Memory and pinned memories | Partial | No | Nothing carries between chats by itself |
| Candy AI | Minimal | Degrades as chats grow | No | Recall weakens through a long conversation |
| SpicyChat | None | No | No | Each conversation is independent |
Verified September 2026 from platform documentation and user reports.
The pattern worth noticing is that memory quality now varies less by category and more by intent. Platforms built around a library of characters treat memory as a convenience, because the next conversation is supposed to start fresh with a different character. Platforms built around one relationship treat memory as the product, because the relationship is meant to deepen. If you are choosing on memory, you are really choosing between those two shapes of product.
What Persistent Memory Costs on Affiny
Memory on Affiny is not sold as a separate feature tier, which matters because the free version has to be usable for the test above to mean anything. Standard memory recall sits on the free tier alongside 300 coins to start, no credit card, and a Daily Gift of 10 to 70 coins. Subscribers get sharper recall and the ability to add, edit, and delete entries in the Memories hub, and viewing what is stored is free on every account.
The coins themselves are separate from memory: chat costs 1 coin per message on the free tier, voice is 30 coins per minute on the free tier and 20 for subscribers, and image generation runs at 15 coins per image free and 10 for subscribers. If memory is the reason you are shopping, the honest advice is to test with the free allowance first and pay only after the recall holds up past day two.
FAQ
What is the difference between in-session memory and persistent memory?
In-session memory is the context the AI has within a single conversation, meaning everything said in the current session. Persistent memory extends beyond the session: when you return the next day or week, the companion remembers previous sessions. Most AI systems have in-session memory by default; persistent memory has to be built and maintained on top of that.
Can an AI companion remember everything I’ve ever told it?
Not literally everything, and not equally. Retrieval-based memory systems store meaningful extractions from conversations rather than full transcripts. The companion remembers what mattered, such as significant facts, emotional context, and things that defined your relationship, rather than a complete verbatim record. This selective retention is actually more like human memory than perfect recall would be.
Why does my AI companion seem to forget things sometimes?
Several reasons: the memory system may retrieve the wrong memories for the current context; a fact may never have been extracted from a prior session in the first place; or the exchange simply did not clear the significance threshold. Some forgetting is also deliberate, because not every minor exchange needs to be recalled forever.
Does memory work across voice calls and text chat?
On platforms with cross-modal memory, yes. Affiny’s memory system bridges both channels: voice call content and text chat content contribute to the same memory pool and are retrievable from either modality. On platforms without cross-modal memory, the companion has separate versions of your relationship depending on which channel you are in.
How does memory affect adult or explicit content?
Memory enables persistent relationship context for adult interactions. The companion remembers your established dynamic, preferred scenarios, and relationship history, which affects how adult scenes are initiated and conducted. Without memory, every explicit interaction starts from scratch. With memory, the companion responds within the context of an established intimate relationship.
What happens to my memories if I switch companions on the same platform?
Memories are typically stored per companion. Switching to a new companion starts fresh, and the memories you built with companion A do not transfer to companion B. On platforms where you build a companion rather than sharing public ones, your memories stay with the companion you built.
The Short Version
Persistent memory is the feature that turns a chat product into something closer to a relationship, and it is the easiest claim in the category to test. Plant three facts, wait a day, and see what comes back without prompting. Platforms that pass carry your identity, your preferences, your shared moments and your emotional context forward. Platforms that fail are running a long conversation and calling it a relationship.
If you want to run the test on a platform built around memory rather than around a character library, you can try Affiny free with 300 coins and no credit card, and check the Memories hub for yourself the next day.