Nectar AI review: custom memory, model tiers and credit costs

Research checked against provider documentation on 26 October 2026. This is an editorial desk review, not a claim of live testing, personal purchases or a UK checkout inspection.

Nectar AI stands out for giving users relatively concrete controls over conversational memory rather than simply promising that a companion "remembers everything". Its documentation describes editable saved memories, automatic memory creation, high-importance stars and tier-dependent context windows. This is attractive if you enjoy writing a detailed fictional world, but model choices, calls and media introduce additional credit decisions.

The memory controls worth paying attention to

Nectar’s memory settings distinguish automatic creation from a switch that allows stored memories to be referenced. You can add an individual custom memory of up to 200 characters. The service lets you star up to ten entries as high importance; the documentation treats the starred set separately from the total number of custom memories. That is a practical way to prioritize a character’s long-running biography or fictional relationships.

Crucially, a new conversation does not inherit the previous conversation’s memories. The docs differentiate Archive, which preserves memories, from Reset, which permanently deletes them. Someone hoping to maintain a long-running storyline should understand those commands before experimenting with the chat settings.

Nectar also describes context windows of 8K for Premium, 16K for Pro and 64K for Ultimate in the published memory guide. Larger context is more text accessible to the model while replying; it is not the same thing as flawless permanent recall. The memory system and the working conversation window should be evaluated separately.

Plan tiers and the price of an action

Nectar’s FAQ says Pro and higher plans allow unlimited companion creation, while Ultimate adds capabilities including long-term memory, video and voice chat. Its help centre also lists credit-consuming actions: some chat models, paid memory options, calls and voice messages beyond allowances, image enhancements and certain video functions. These charges can change, so a fixed promise that all calls or images are included would be unsafe.

To evaluate a plan, start with the activity you intend to repeat. For example, a user who creates characters but rarely generates media may care about the editing and context limits more than a video credit allowance. Someone who wants daily calls should investigate the actual rate per minute before upgrading. Compare the plan’s recurring price and credit bundle, then estimate a normal week of paid actions. We found no verified UK checkout receipt, so we avoid inventing a sterling total.

A reproducible test without private information

Create a fictional character called Alex who works night shifts at a lighthouse, dislikes thunderstorms and drinks mint tea. Save each detail separately as permitted. Start a story about a weekend invitation, then ask when to visit; this tests whether the schedule is integrated into an answer. Introduce a contradictory fact later and edit the saved memory to see whether the service still refers to the old story. Finally, archive rather than reset the conversation and check what is retained. Keep a note of the plan, model and whether automatic memory was enabled to avoid comparing different settings.

UK purchasing and our conclusion

The site is browser-based and its publicly described premium features do not verify a GBP-inclusive final price. UK readers should check the currency, applicable tax, cancellation route and recurring charge at the checkout. Also confirm whether paid credit balances continue after a downgrade. Nectar is more compelling when manual memory management is the problem you want solved than when you primarily want to browse polished premade characters. Consider Nomi for a conversation-led approach with less tinkering, or Kindroid for further character-configuration options.

Sources and research limits

• Nectar — memory documentation
• Nectar — FAQs, plans and paid features
• Nectar — official feature overview
• Nectar — troubleshooting and memory reset