Nectar AI: features and setup guides

Service profile

Nectar AI

The evidence supports French chat, companion customization and memory-related plan differences; relationship-level customization and Ultimate’s 64K context are provider editorial claims without precise UI documentation. Editable memory, character-card import, dedicated pronoun controls and explicit Friend/Romantic settings remain unverified, not necessarily unavailable. Advisory pronoun fallback: write, 'For this chat, refer to me as [name] and use [pronouns].' This is an identity instruction, not a dedicated control or a guarantee of persistence. Advisory Japanese fallback: write, 'Please reply in Japanese for this chat.' Japanese support, persistence, voice output and tier eligibility remain unverified; receiving a Japanese reply would not verify those capabilities. If stored-memory editing or automatic card import is essential to your purchase, this documentation does not establish that Nectar meets those needs. No application behavior was tested.

Editable or persistent memory
Documented context / conversation window
Documented French chat setup
User persona / profile controls

Checked 2026-10-02 · 8 sources

See current Nectar AI plans Compare with your must-haves

What it can do

Capabilities we could verify

Editable or persistent memory: Use companion chat with persistent memory, subject to technically limited recall. Paid plans extend memory, but no numerical limits or memory editor are documented. After a subscription ends, premium memory becomes unavailable, premium-tier history may become inaccessible, and free-plan memory limits apply. Advisory fallback: supply a short current-chat recap of important facts and corrections. This is not viewing, editing or deleting stored memories, and does not guarantee persistence.
Documented context / conversation window: Nectar’s comparison article says Ultimate provides 64K context memory and separately lists long-term memory. This is a plan-specific editorial claim, not independently corroborated pricing documentation or measured performance. It does not establish editable memory or guarantee that particular details will be recalled.
Documented French chat setup: Chat in French: the FAQ explicitly lists French as supported. No selector, setup path, required prompt, model requirement, persistence behavior or tier restriction is established. Advisory text-chat fallback: write, 'Please reply in French for this chat.' This is a suggested instruction, not a documented setup procedure; French voice output is not verified by this evidence.
User persona / profile controls: Create a companion by customizing personality, backstory and speaking style. Advisory alternative to unverified character-card import: use the card’s contents as a reference and manually recreate those documented aspects. This does not imply card uploading, automatic field mapping or support for any file format. No eligible tier is specified. The Terms prohibit image uploads during character creation.
Romantic relationship setting: The product-specific setup article says users can set romance, intimacy and dependency levels after choosing or customizing a character. It does not identify control names, available values, saving steps, later-editing procedures or eligible tiers. This supports a broad relationship-customization claim, not a verified dedicated Romantic setting or a Friend mode that constrains romantic responses.

Before you pay

Check which plan actually contains the feature you need, the billing period, credit or usage limits, cancellation terms, and whether text, voice and interface-language support are separate.

Useful before switching

Keep a portable setup first.

Save the character description, relationship, continuity notes and opening message before opening or changing an account.

Build a reusable setup

Guides that actually apply to Nectar AI

No provider-specific guide has passed the current content checks yet.

Sources

Checked 2026-10-02

Provider documentation tells us what a service describes. Unless a page explicitly says otherwise, we are not presenting it as a hands-on performance test.