Secrets AI Features: Everything the Platform Offers in 2026

You have probably seen AI companion platforms that offer decent conversation and little else. Secrets AI was built around a different philosophy: layer conversation, image generation, video clips, voice, and persistent memory into a single environment. The result is a feature set that is genuinely broader than most competitors in this space. This page maps every capability the platform offers — from its five AI chat models down to the edge cases like group chats and identity personas — so you can see exactly what you are getting before you commit.

For context on pricing per feature, the detailed pricing page shows Moments costs for each action. The full review gives an overall assessment and verdict.

The Conversation Engine: AI Chat Models and How They Work

At the center of Secrets AI is a chatbot architecture powered by five selectable AI models. The technical differences between these models are not publicly documented, but in practice they produce meaningfully different conversation styles — varying in emotional nuance, response complexity, and tone. Premium AI models, available from the Premium tier upward, consistently produce more sophisticated and context-sensitive replies than the base models.

The conversation system is built on natural language processing that prioritizes companion dynamics over factual Q&A. The AI does not browse the internet or answer productivity questions — it is optimized for emotional engagement, roleplay, and relationship simulation. Conversation starters are provided for every character: roughly half are casual openers, the other half are more suggestive, lowering the barrier to getting a conversation moving.

Roleplay support is built in from the start. Pre-configured scenarios include office romance, teacher-student dynamics, landlord interactions, boss scenarios, and a range of fantasy contexts. The AI handles scene transitions and narrative pacing well — reviewers note coherent storytelling and vivid descriptive writing within extended roleplay threads.

The Gallery: Character Library and Customization

Pre-Built Characters

The character library contains approximately 94 pre-built companions spanning realistic and anime visual styles. Each character comes with a defined personality and visual profile. Access to the full library is available even on the free tier, meaning you can browse and evaluate characters before spending anything.

Paired with the character library are 100+ fantasy scenarios that determine the initial relationship context. Character diversity is a documented limitation — reviewer scores land at 3.8/5, noting that while scenario variety is broad, the underlying character designs skew toward a narrower range of ethnicities and body types than the scenario count would suggest.

Custom Character Creation

Custom character creation gives you direct control over your companion's appearance and personality. The first custom character is free; additional characters cost 1,000–1,500 Moments.

Customization parameters include:

• Appearance: Ethnicity (9 options), skin tone, eye color, hairstyle, body type (6 options)
• Voice: 6 voice options, each with its own natural tone
• Personality traits: Confidence, playfulness, dominance, intelligence (adjustable sliders)
• Background: Profession (5 presets), personality archetypes (5 presets), hobbies, and a free-text field for specific details
• Scenario context: Set the relationship dynamic from the start

The free-text field is particularly useful — entering specific preferences or backstory details here produces noticeably more personalized responses compared to using only the preset options.

The Visual Layer: Image Generation

Image generation runs on two selectable models — a standard model and an Advanced/Premium model available on higher tiers. Four image presets are available, covering different lighting styles and framing approaches.

Key facts about image generation:

• Cost: 25–50 Moments per image
• When you create a new character, 4 images are automatically generated (included at no extra cost)
• Group image generation is supported — multiple characters in a single frame
• Quality is generally high with occasional anatomical glitches on hands and complex poses
• Realistic rendering is the default style; the anime characters maintain their distinct visual treatment

The video generation feature extends from image generation — you select an existing companion image and add a motion prompt to produce a video clip.

The Motion Layer: Video Generation

Video generation is Secrets AI's most distinctive capability. Most competing platforms in the AI companion space — including CrushOn AI, Janitor AI, and Character.AI — do not offer this. Candy AI offers limited video; Secrets AI's implementation is more developed.

The workflow is straightforward: select or generate a companion image, write a prompt describing the desired motion or action, and wait approximately 2 minutes for the clip to render. The AI applies the motion to the character's existing appearance and scenario context.

Cost range: 50–600 Moments per clip, varying by clip length and quality settings. Short 3-second clips cost approximately 50 Moments; longer, higher-quality clips approach 600 Moments. Video quality is rated 4.1/5 — described as natural movement and realistic expressions, with occasional variation on complex prompts.

This feature is available from the Lite tier upward; free users cannot access video generation. For the complete cost breakdown by tier, see the video generator deep dive.

The Voice Layer: Voice Calls

Six voice options cover a range of tones and characteristics. Voice call quality is rated 4.3/5 — noted for natural adaptability to conversation context rather than flat, robotic delivery.

Voice calls cost 100 Moments per minute, which makes them the second most Moments-intensive feature after video. On the Premium tier (8,000 Moments/month), budget allocation translates to roughly 80 minutes of voice time — or a mix of voice plus other media types.

The Memory Architecture

Memory is where the biggest functional gap between free and paid exists.

The emotional timeline (Premium+) means the AI tracks the emotional arc of your relationship over time — recognizing shifts in tone, recalling significant moments, and adjusting its behavior based on the accumulated context.

Time Machine (Premium and above) allows you to revisit and replay past conversation states — a feature designed for users who want to explore alternative narrative directions within an existing relationship.

Manual memory saves cost 10 Moments per entry. These let you pin specific facts or preferences so the AI retains them regardless of the automatic memory window.

Context length boosts — +60% on Premium and +100% on Ultimate — affect how much of the current conversation the AI actively references when generating each response, producing more coherent extended threads.

The Advanced Layer: Group Chats and Personas

Both of these features are exclusive to the Ultimate tier ($39.99/month).

Group chats allow up to 3 AI characters to participate in a single conversation simultaneously. Group image generation is supported alongside this — you can generate images featuring multiple characters in the same scene.

Personas enable identity-switching during roleplay: you can define alternative identities for yourself within conversations, allowing for more complex roleplay structures where your character's role shifts dynamically.

The Security Layer: Privacy and Safety Features

Private Mode applies end-to-end encryption to a conversation on a per-session basis. Enabling it also opts that specific conversation out of AI training data collection. This must be toggled on per conversation — there is no global default setting. Private Mode is available on all tiers.

Account privacy by default includes:

• Anonymous signup (email and password only — no real name, phone number, or social login required)
• Billing under the descriptor "Sun Clinical Laboratories" (no reference to AI or adult content)
• Account deletion that removes all data within 30 days

Information privacy limitations worth noting: the platform does not publish specific details about encryption protocols or data retention timelines. Independent reviewers rate privacy documentation at 2.9/5. The safety analysis page covers what is and is not documented.

The Gaps: What Secrets AI Does Not Offer

No platform covers everything, and being clear about limitations is part of accurate evaluation. As of 2026:

• No native mobile app (browser-only; no iOS App Store or Google Play listing)
• No push notifications
• No offline access
• No biometric login (fingerprint or face recognition)
• No multi-modal input (you cannot send images to the AI — generation flows only from AI to you)
• Limited preset variety for images and videos
• Character diversity narrower than the scenario count implies

For users who need a native mobile experience, the alternatives comparison covers PocketGirl AI and other mobile-first options.

FAQ