SpicyChat Review
SpicyChat suits adults who want fast, free text roleplay with a huge catalogue of community characters and a private character creator; replies arrived in about 3.5 seconds and we hit no message cap. The serious catch is moderation: the platform itself offers an "Age Play" tag and non-consent tags as adult filters and promoted school-themed and "just turned 18" cards in Trending, despite its own bans on minors and non-consent, while age checks in many countries are only self-confirmation. Free-tier memory was also weak, recalling 1 of 4 story facts.
Score breakdown
Our 5.8 comes from a 100-point checklist we use for every tool: 58 / 100. Tap a row to see why.
Features & functionality11 / 18
Free chat worked for 38 messages with no cap, queue or failed generation; replies firmly in character; regenerate (~5-8 s) and edit work; private character creator verified (3 messages); image attachment understood on free; Web Search and Command (OOC) in the + menu; 6 free and 16 paid models. Deductions: memory recalled 1 of 4 planted facts with invented substitutes (4K context, 180-token replies), edit does not regenerate, no message delete or rewind, no chat export, image generation and voice paywalled, a reply cut mid-sentence.
Signup & ease of use7 / 12
Passwordless email code arrived immediately; username, display name and 18+ checkbox only. Stuck point: header still showed 'Sign In' after onboarding and a blurred Community Guidelines modal blocked clicks until accepted plus two reloads; ~10 s redirect to Kinde and 10-20 s 'Loading...'; interest-picker pop-up on first visit; explicit content enabled from the profile; 'Context Limit Reached' upgrade toast after every reply from message 14.
Reliability8 / 12
No crash, queue, rate limit or failed send in 38 messages; data persisted. Defects: stuck signed-out header after onboarding, one reply cut mid-sentence, garbled user name once, invented memory details, a 'Directed:' OOC instruction answered the previous question, edit without regeneration.
Design5 / 8
Consistent dark UI with clear cards, tags and counts; detailed creator form. Against: repeated upgrade toasts and rotating banners, 'Feature release' and interest pop-ups, generic icon-named controls, 'Im All In' typo on the plans page; Lighthouse CLS 0.578 desktop / 0.747 mobile (visible layout shift).
Speed4 / 8
Chat fast: 2.4-9.1 s per reply, median ~3.5 s (12 timed, some estimated), regenerate ~5-8 s, but no streaming. Page load weak: browser HTML ~2.5 s, app visible ~6-8 s; Lighthouse performance 54 desktop (LCP 3.0 s, CLS 0.578) and 19 mobile (LCP 12.7 s, TBT 1,610 ms, TTI 13.5 s, CLS 0.747).
Mobile4 / 8
500px window: hamburger menu and stacked composer work, but a 'Feature release' pop-up covered the chat header and menu. Lighthouse mobile 19/92/92/100. Android only as an APK from the website (not on Google Play); iOS app removed from the App Store (company docs suggest Add to Home Screen). Apps not tested.
Payments & billing5 / 8
Clear USD prices on the plans page matching checkout ($5.95/$19.95/$29.95 monthly; $60/$199.95/$299.95 yearly); embedded Sticky.io checkout with card (Visa, Discover, Mastercard), one-time Alipay and crypto; renewal and advance-notice text, descriptor shown, unticked terms checkbox, coupon and plan switch, cancel from profile. Deductions: USD only with no tax line, no PayPal or wallets, refund rules contradictory and restrictive, chargebacks trigger an automatic ban (help article), descriptor differs from help. Not paid.
Security & privacy3 / 7
Passwordless email codes via Kinde, CSP in enforcement, X-Frame-Options, nosniff, DNSSEC, DMARC p=quarantine with DKIM, permanent account deletion offered. Missing: 2FA, session list, data export, HSTS on the app HTML; cookies without Secure/HttpOnly; Google Analytics and Ads fire with no consent banner; Privacy Policy dated 4 May 2023 with under-13 wording and no statement on chat training; a third-party write-up reports a patched July 2026 message-access flaw.
Company & trust4 / 7
NextDay AI Incorporated verified active on Corporations Canada (no. 1516881-3, incorporated 2023-07-04, annual filings 2024-2026, office matches footer); domain registered 2023-05-01 with DNSSEC; public docs, help centre and Discord. Against: Trustpilot 2/5 'Poor' (capture), iOS app removed, Android outside Google Play, Delaware virtual-office address and Cyprus entity unverified, stale Privacy Policy, no status page, and a catalogue that surfaces content its own rules ban.
Accessibility2 / 5
Lighthouse a11y 96 desktop / 92 mobile and chat buttons show a focus ring. Keyboard failures: no skip link, every sidebar item takes 2 tab stops (link named 'item' wrapping a button), no visible focus in the sidebar after 25 tabs, chat input labelled only by placeholder, buttons named by icon ids; Lighthouse label-content-name-mismatch and target-size failures. No screen-reader test.
Honest content2 / 4
Extensive docs, help centre, community guidelines and a 2257 page. Contradictions: guidelines ban non-consent and minors while the platform offers CNC and Age Play tags; I'm All In context 16,384 (help) vs 32,768 (docs, plans page); refund rules 24 h (help) vs 'exceptional circumstances' (FAQ) vs non-refundable once logged in (/refund); descriptor NDAI.CC (help) vs NDAI.CHAT (checkout); Terms vs Guidelines on public-figure bots; 'NexDay' typo in the Terms.
Findability3 / 3
Lighthouse SEO 100 desktop and mobile; robots.txt allows all; sitemap index with 60 child sitemaps (50,000 character URLs in the first alone), 810 tag URLs and 9 language prefixes; title, meta description and canonical present. Minor: legal pages not in the static sitemap; curl receives an empty SPA shell.
Pros & cons
What we liked
- Fast replies, median about 3.5 seconds
- No message cap in 38 free messages
- Private character creator with lorebook support
- Free plan understands attached images
- Named Canadian company with active registry record
What could be better
- Offers age-play and non-consent tags with NSFW
- Recalled only 1 of 4 story facts
- Upgrade toast after every reply past message 14
- Self-declared age check in many countries
- No 2FA, chat export or message delete
What went wrong in our test
22 problems found: 2 critical, 2 high, 10 medium, 8 low.
- criticalPlatform offers an Age Play tag with adult content and promotes school-themed and 'just turned 18' cards
An 'Age Play' tag is offered as a filter (78,000 cards with explicit content on), and the top ~30 Trending cards included a 'schoolgirl (18+)', a 'just turned 18' and a 'freshman' card and an Age Play + NSFW card; not opened.
Where: Home > Trending (explicit content on); tag filter - criticalNon-consent themes offered as platform tags and surfaced in Trending
CNC (32,309), Slave (26,212) and 'free use' cards are offered and promoted although the guidelines ban non-consensual actions 'even in fictional settings'; not opened.
Where: Tag filter; Home > Trending (explicit content on) - highUnexplained 'Real' tag paired with adult content
A 'Real' tag (40,300 cards with explicit content on) is offered without definition; teen-framed Trending cards carry it, and its meaning (realistic style or real person) is unclear.
Where: Tag filter; Trending cards - highSelf-declared age only, no age gate for visitors
No age gate or interstitial before the catalogue with suggestive thumbnails; 18+ is a checkbox at onboarding and a self-confirmation in the profile; document or selfie checks only in listed regions (company claim).
Where: Homepage (logged out), onboarding, Profile > Display explicit content - mediumFree-tier memory recalled 1 of 4 facts and invented the rest
13-16 messages after planting, only the name was recalled; scarf, dog name and fear were replaced with invented details.
Where: Chat (free, 4K context, Default model) - mediumUpgrade toast after every reply once context is full
A 'Context Limit Reached ... Upgrade for 4x longer conversations' toast reappeared after every reply from message 14, plus rotating upsell banners on Chats.
Where: Chat (from message 14); Chats page - mediumSession stuck signed-out after onboarding behind a blocking modal
After Continue the header still said 'Sign In' even after reload, and a blurred guidelines modal blocked clicks until accepted and the page reloaded again.
Where: Post-signup redirect, header, Community Guidelines modal - mediumNo 2FA, session management or data export
Accounts holding private adult chats rely on email codes or social login only; no 2FA, no active-session list and no personal-data export.
Where: Profile settings - mediumNo chat export and no message delete or rewind
Messages can be regenerated or edited but not deleted or rewound, and chats can only be removed, not exported.
Where: Chat message menu; Chats page - mediumAnalytics and ad tracking without consent; stale Privacy Policy
Google Analytics and Google Ads requests fire with no consent banner; the Privacy Policy dates from 4 May 2023, predates the company, uses under-13 wording and does not address chat training.
Where: All pages; /privacy - mediumContradictory and restrictive refund rules
Help allows refunds within 24 h, the FAQ says 'only under exceptional circumstances', /refund (7 Oct 2026) says non-refundable once logged in, and a payment dispute triggers an automatic ban.
Where: Help 'Refunds', FAQ, /refund, help 'Account shows blacklisted' - mediumKeyboard and labelling failures
No skip link, double tab stops on a link named 'item', no visible sidebar focus, placeholder-only chat input and buttons named by icon ids.
Where: Sidebar navigation, chat composer - mediumPoor mobile page performance and layout shift
Mobile performance 19 with LCP 12.7 s, TBT 1,610 ms, TTI 13.5 s and CLS 0.747; desktop 54 with CLS 0.578.
Where: Homepage (Lighthouse mobile and desktop) - mediumCharacter creator has no content rating or on-form safety rules
The form has a guidelines checkbox but no content-rating field and no text on minors, real people or consent, while Public publishing is one option away.
Where: /chatbot/create - lowBilling descriptor differs between help and checkout
Checkout says purchases appear as NDAI.CHAT; the help centre says NDAI.CC.
Where: Checkout summary vs help 'Payment methods' - lowUSD-only checkout with no tax line
The total was identical with Turkey or US selected and no tax line was shown, so buyers cannot tell whether tax or conversion fees will be added.
Where: /checkout - lowOfficial context figures disagree
I'm All In context is 16,384 tokens in help but up to 32,768 in docs and on the plan-comparison page.
Where: Help 'Plans and what's included' vs docs and plans page - lowReply cut mid-sentence; no streaming
A retest reply ended mid-sentence, and replies appear whole after the wait instead of streaming.
Where: Chat (free, 180-token cap) - lowModel slips on names, OOC commands and platform claims
The private test character garbled the user's name once, a 'Directed:' instruction answered the previous question, and the model claimed the platform requires non-explicit content, which the site does not say.
Where: Chat output (free models) - lowFeature-release pop-up covers the chat header at 500px
A 'Feature release' pop-up overlapped the chat header and menu in a narrow window.
Where: Chat at 500px width - lowNo HSTS and cookies without Secure flag
The app HTML has no Strict-Transport-Security header, and country/region cookies are set without Secure or HttpOnly.
Where: https://spicychat.ai/ response headers - lowLegal-text inconsistencies
The Terms spell the US subsidiary 'NexDay AI USA Inc', the Privacy Policy gives only 'Corporate Address' and predates the incorporation, and it addresses under-13s on an 18+ service.
Where: /terms, /privacy
What it is
SpicyChat is a web platform for text roleplay with AI characters, run by NextDay AI, a company registered in Montreal, Canada. Almost every character is made by another user. You browse a public catalogue of hundreds of thousands of cards, open one and chat, or build your own with a personality, scenario, example dialogue and an optional lorebook. The free plan includes text chat on six models with a 4K-token memory window. Three paid tiers add longer memory, image generation in chat, voice replies and more models. It is meant for adults only. There is an Android app downloaded from the website, but no current iPhone app.
What it’s like to use
There is no age gate or cookie notice on the homepage, and the catalogue loaded in about six to eight seconds. Sign-up is passwordless: you enter an email and type a six-digit code, which reached us immediately. Google, Apple, X and Discord sign-in are also offered. Onboarding asks for a username and an 18+ checkbox. We got stuck once: after onboarding, the header still showed “Sign In” and a blurred guidelines window blocked every click until we accepted it and reloaded twice.
Adult cards stay hidden until you switch on explicit content in your profile, which only asks you to confirm you are over 18. With it on, the Trending count jumped from about 385,000 to about 1.45 million cards.
We sent 32 messages to a popular fantasy character, 3 to a private character we made, and 3 to an adult character in a separate test. Replies were fast, between 2.4 and 9.1 seconds with a median of about 3.5, though text appears all at once instead of streaming. We hit no message cap, queue or cooldown. The writing stayed firmly in character, and regenerating a reply took about five to eight seconds. Attaching a picture worked on the free plan, and the bot described it correctly.
Memory was the weak spot. We planted four facts early on, and 13 to 16 messages later the character recalled only our name; it invented the other three. From message 14, a “Context Limit Reached” notice offering an upgrade appeared after every reply.
The character creator is detailed, with long text fields, up to 12 tags and a private setting by default.
Where it falls short
Moderation is the biggest concern. The site’s rules ban minors and non-consensual content, yet the platform itself offers an “Age Play” tag and non-consent, “free use” and slave tags as filters once adult content is on. Its Trending list showed school-themed and “just turned 18” cards near the top. A “Real” tag is offered as well, and its meaning is not explained. We did not open any of these cards. The creator form has no content rating and no reminder of the rules on minors, real people or consent. Outside the regions where the company says it runs selfie or ID checks, age assurance is a checkbox.
We could not confirm how explicit the free model gets. Both adult test chats ended with the character slowing things down in character, and no platform filter message appeared.
Account tools are thin: no two-factor login, no session list, no data or chat export, and no way to delete or rewind a single message. Editing a message does not regenerate the reply.
Billing is in US dollars only, and the checkout showed no tax line. The payment page lists a different bank-statement name from the help centre, and the refund pages disagree, with the newest saying payments are non-refundable once you log in. Keyboard users get no skip link and unlabelled controls. In a 500-pixel-wide window, a pop-up covered the chat header, and lab tests gave the homepage a mobile performance score of 19 out of 100.
Who it’s for
SpicyChat suits adults who want quick, free text roleplay with a very large choice of community characters and a private creator, and who can live with short memory unless they pay. Skip it if you want a carefully moderated catalogue, strong account security, or confirmed explicit output before you subscribe.
Full test notes
Everything we did and measured, step by step.
Signup & Onboarding
- Finding signup: "Sign In" in the header opens a modal: email code, Google, Apple, X or Discord.
/registerreturns "Page Not Found"; no separate signup route. - Bot protection: none visible (no captcha or challenge).
- Verification: ~10 s redirect to
auth.spicychat.ai(Kinde). Email from "NextDay AI Roleplay", subject "Email verification code", 6-digit code valid 2 hours, arrived immediately. Return redirect showed "Loading..." for 10–20 s. - Onboarding: "Welcome Aboard" modal — username (20), display name (20), "I confirm that I am 18 years or older" checkbox with Terms/Privacy links. No tour. An "I'm interested in..." gender picker appears on first visit.
- Stuck point: after Continue, the header still said "Sign In" even after a reload, and a blurred "Community Guidelines Agreement" modal blocked clicks. Accepting it and reloading again fixed the session.
- Age: no visitor age gate; checkbox at onboarding; adult cards require "Display explicit content - Confirm that you are over 18" in the profile. Video-selfie/ID checks only in France, the UK, Italy, Australia and 26 US states (company claim). .
- First meaningful result: first reply within seconds once the session worked.
- Payment details: never requested on Free. "Free" badge; no message or credit counter.
Feature-by-Feature Analysis
Catalogue — works, poorly moderated. NSFW off by default; enabling it raises Trending from 385,518 to 1,446,551 results. Tags with counts (explicit on): NSFW 811,300; CNC 32,309; Slave 26,212; NTR; Vore; "Age Play" 10,200 (off) / 78,000 (on); "Real" 8,700 / 40,300. Flagged cards in the top ~30 Trending (explicit on), not opened: a card described as a "naughty schoolgirl (18+)"; a "neighbor who just turned 18" (tagged Real); a card tagged Age Play + NSFW; a "freshman" card (tagged Real); a "free use" card; a Slave-tagged card. With explicit off: a "school bully" card and an "all boys school" card. The profile "Highlights" placeholder example describes a "petite and slim 22 years old girl ... like a princess". A creators leaderboard awards a free month of the top tier, which rewards whatever trends. See , -002, -003.
Character creator — verified. Avatar, name (20), title (100), greeting (2,000), personality (5,000), card import (.json/.png), visibility Public/Private/Unlisted (Private default), up to 12 tags, lorebook, rulebook, voice (I'm All In), scenario (4,000), example dialogue (4,000), groups, 2,400-token limit, guidelines checkbox. No content-rating field and no on-form rules about minors, real people or consent. A private adult original character was created and chatted with for 3 messages: in character, garbled the user's name once, 2.4 s reply.
Chat — works, short memory. Character: a popular fantasy knight-commander (115.5k chats; no age stated on the card). User character: an injured traveller, stated to be an adult. Model: free Default (26B).
- Settings (free): 4K context, 180-token max reply, temperature 0.7, Top-P paid-only; 6 free models, 16 paid.
- Speed: 2.4–9.1 s, median ~3.5 s over 12 timed replies (some estimated); no streaming; regenerate ~5–8 s.
- Memory: of 4 planted facts (name, a red scarf from the user's mother, a white dog's name, fear of deep water), only the name was recalled 13–16 messages later; the other three were replaced with invented details. The context-limit toast had appeared at message 14.
- Staying in character: very consistent. "(OOC: ...)" was converted into a "Directed:" message; the in-story AI question was deflected, and a later directed message produced "I am exactly that" (a "construct") while answering the previous question rather than the instruction.
- Controls: regenerate works (2/2 variants); edit saves (10,000-char limit) without regenerating; no delete or rewind.
- Explicitness: inconclusive. The main character refused romance in character (tester error noted in §3); the retest character consented to a kiss and then slowed down in character, with the third reply cut mid-sentence. No platform warning, filter banner or moderation notice at any point. An OOC answer claiming the platform "generally requires" non-explicit content is a model statement, not platform policy.
- Limits: no cap, cooldown, queue or in-chat ads in 38 messages. From message 14 a "Context Limit Reached ... Upgrade for 4x longer conversations" toast after every reply; rotating banners on Chats.
- Media: image attachment works on free (a 512px test image was described correctly, 7.7 s); image generation (True Supporter+) and voice playback (I'm All In) open upgrade modals. "+" menu: Attach Image, Web Search, Command (OOC). Web Search was not exercised.
Account. Username, name, plan, avatar generate/upload, highlights, default model, language; Advanced > Remove Account (permanent, not used). No 2FA, session list, password or data export. Chats removable singly or in bulk; no export.
Content policy (published). Guidelines: all users and characters 18+; "No bot should depict, promote, or involve in any non-consensual actions, even in fictional settings"; incest and step-family roleplay banned; aging up of characters from teen-oriented media allowed if explicitly 18+; youthful demeanours "scrutinized individually". Terms ban sexualised minors including fictional depictions, non-consensual sexual activity, and sexual content of real identifiable people. The live tag taxonomy and Trending list are inconsistent with these rules.
Technical Quality
Reliability. No crash, failed generation or rate limit in 38 sends. Defects: (stuck session), -018 (cut reply), -019 (model slips), edit without regeneration.
Performance.
| Lighthouse 13.6.0 (homepage) | Perf | A11y | Best pr. | SEO | FCP | LCP | TBT | CLS | SI | TTI | Weight |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Desktop (05:10:07) | 54 | 96 | 92 | 100 | 0.5 s | 3.0 s | 140 ms | 0.578 | 2.3 s | 3.1 s | 1,725 KiB |
| Mobile (05:10:40) | 19 | 92 | 92 | 100 | 1.7 s | 12.7 s | 1,610 ms | 0.747 | 6.1 s | 13.5 s | 1,712 KiB |
Browser session: HTML ~2.5 s, catalogue visible ~6–8 s. Chat replies fast (§5). Best Practices lost points to a Cloudflare Insights beacon blocked by the site's own CSP and missing source maps.
Mobile. 500px window: hamburger menu and stacked composer work; a "Feature release" pop-up covered the chat header and menu. Android: official APK from the website. The same operator's SFW app PixelChat is on Google Play (Teen, ~3.2/5, 500K+ downloads). iOS: official app removed from the App Store; docs recommend Add to Home Screen; a lookalike app from a different seller exists. Not tested on a phone.
Security (observable). HTTPS with TLS 1.3 and a Google Trust Services certificate; CSP enforced (with unsafe-inline), frame-ancestors 'self', X-Frame-Options SAMEORIGIN, nosniff, referrer-policy: same-origin. No HSTS on the app HTML; country/region cookies without Secure/HttpOnly. DNS: DNSSEC signed, DMARC p=quarantine strict alignment, DKIM, SPF ~all, CAA. Auth via Kinde with email codes; no 2FA. A third-party bug-bounty write-up (7 July 2026) reports a patched message-labelling authorization flaw; not retested and not treated as open.
Privacy (company claims + observations). Privacy Policy last updated 4 May 2023: collects email, names, usage/device data, cookies and social-login data; chat content not clearly listed; no GDPR/CCPA section; under-13 wording. Terms (23 July 2026) add a retention policy and a broad licence to use and sell user-submitted content. Google Analytics, Google Ads, Firebase, Recombee, TrackJS and Tapfiliate load with no consent banner.
Pricing & Payments
| Plan | Monthly | Annual | Checkout checked | Notes |
|---|---|---|---|---|
| Free | $0 | — | — | 4K context, 180-token replies, 6 models; no cap in 38 messages |
| Get a Taste | $5.95 | $60.00 (≈$5.00/mo) | Monthly + Annual | "Chat Without Ads", key-detail memory, 10 personas; marked best: true in File A as the lowest monthly-equivalent price |
| True Supporter | $19.95 | $199.95 | Monthly | "Most Popular"; 2x memory, in-chat images, longer replies, 50 personas |
| I'm All In | $29.95 | $299.95 | — (plans page) | 8x memory, best models, voice, 100 personas |
- Processor: embedded checkout on
spicychat.ai/checkout, card fields fromcdn.sticky.io(Sticky.io), fraud check fromssl.kaptcha.com. "Payments are secure and encrypted" (company claim). - Methods seen: credit card (Visa, Discover, Mastercard), Alipay (one-time, ¥), cryptocurrencies (one-time). Apply Coupon; Change Plan. No PayPal, Apple Pay, Google Pay or Pay by Bank at checkout; docs also mention Pay by Bank, Boosty, CoinPayments, SubscribeStar and Google Play via PixelChat.
- Currency/tax: USD only even with Turkey selected; same total for Turkey and US; no tax line.
- Required data: first/last name, country, province, postal code; email prefilled. Your purchase will appear as NDAI.CHAT on your statement and you can cancel anytime from your profile." Terms checkbox unticked by default. Help says the descriptor is NDAI.CC.
- Refunds: help article — within 24 h with minimal usage (card); FAQ — "only under exceptional circumstances";
/refund(updated 7 Oct 2026) — non-refundable once the login is used, no partial months, approved refunds within 10 days to the original method, crypto excluded. Chargebacks lead to an automatic ban per help. - Cancellation (company claim): Profile > Manage Subscription → payment platform; access continues to period end. Pay by Bank has no Manage Subscription; Boosty and crypto are linked manually; Play/SubscribeStar cancel on those platforms; deleting the account does not cancel third-party subscriptions.
- Not paid; the account remained on Free.
Company Credibility
| Item | Status | Finding |
|---|---|---|
| Legal entity | Verified (primary registry) | NextDay AI Incorporated / ProchainJour IA Incorporée, Corporations Canada no. 1516881-3, active, incorporated 2023-07-04, annual filings 2024–2026, office 200-4388 Rue Saint-Denis, Montréal (matches footer) |
| Subsidiaries | Company claim | NextDay AI USA Inc (2915 Ogletown Road, Suite 4642, Delaware — a common registered-agent/virtual-office address) and NextDay AI EU Ltd (2 Poreias, Limassol); not re-checked in registries; "NexDay" typo in the Terms |
| Governing law | Company claim | Quebec law, Montreal courts, individual arbitration and class-action waiver |
| Domain | Verified | Registered 2023-05-01 (Cloudflare Registrar), expires 2029, DNSSEC; ~3.5 years old |
| Apps | Verified | Android APK only from the website; iOS app removed; PixelChat (SFW, same operator) on Google Play |
| Reputation | Mixed, anecdotal | Trustpilot "Poor", 2/5 (capture); Reddit themes: large catalogue and usable free tier praised; complaints about memory, surprise filtering, age verification and the September 2026 price rise |
| Breaches / lawsuits | None found | No primary report found; absence is not proof |
| Incidents | Third-party, unverified | Outage reports (Oct 2025, 9 Nov 2026); no first-party status page |
| Moderation record | Warning sign | Platform tags and Trending surface content its own rules ban (, -002) |
Overall, SpicyChat's ownership is more transparent than many adult-AI services (primary registry record, matching address, consistent filings), but the stale Privacy Policy, missing iOS presence, weak public reputation and the gap between published rules and the live catalogue reduce trust.
How we tested
We used SpicyChat without a paid plan for 1 days: 1 day, 1 account, 3 characters (1 self-made private), 38 messages on the free plan, 1 image attached, 1 regenerate, 1 edit; 3 of 6 paid plan options checked up to checkout without paying. We scored it against the same checklist we use for every tool in Roleplay Chat.
- ✓Created an account
- ✕Went through checkout without paying
- ◷Tested on Nov 13, 2026
What we could not test (11)
- No purchase made: 3 of 6 plan/period combinations (Get a Taste monthly and annual, True Supporter monthly) were taken to the embedded checkout and stopped before confirmation; paid features (longer context, Memory Manager, Semantic Memory, in-chat images, voice, paid models) were not verified
- Single test day, one account; 38 messages on the free plan across 3 characters (32 to one popular character, 3 to a self-made private character, 3 in an explicitness retest); chat-quality findings are indicative
- The main test character's card states no age; romance testing used an adult user persona only
- Tester error: in the romance test two further escalating messages were sent after the character's first in-character refusal; testing stopped after the third refusal and the next message de-escalated
- Explicitness not established: both adult test chats ended with in-character slow-downs and no platform filter message, so the uncensored sub-score is omitted
- Flagged catalogue cards (Age Play, school or 'just turned 18' framings, Real, CNC, free use, Slave) were seen in listings only and were not opened, so their content was not verified
- Mobile judged from a 500px-wide desktop window and Lighthouse mobile emulation; the Android APK was not installed and no real phone was used; no current iOS app
- Accessibility: keyboard pass and Lighthouse only; no screen-reader test
- Account deletion, logout/login cycle, support response, refunds, regional age verification and social logins were not tested
- Delaware and Cyprus entities not re-checked in official registries; Trustpilot rating from a fetch-provider capture (direct request returned HTTP 403)
- Scored independently; no comparison with any product owned by ExplicitList's owner
Pricing
| Plan | Price | Good for |
|---|---|---|
| Free | Free | 4K context, no cap seen |
| Get a Taste Monthly | $5.95/mo | Memory tools, no ads |
| Get a Taste AnnualBest value | $60/yr | About $5 a month |
| True Supporter Monthly | $19.95/mo | 8K context, chat images |
| True Supporter Annual | $199.95/yr | About $16.66 a month |
| I'm All In Monthly | $29.95/mo | Voice, all models |
| I'm All In Annual | $299.95/yr | About $25 a month |
FAQ
Is SpicyChat free?
Yes. We sent 38 messages on the free plan without hitting a message cap or queue, but the free context window is 4K tokens and an upgrade notice appeared after every reply from message 14.
How much does SpicyChat cost?
Get a Taste is $5.95 a month, True Supporter $19.95 and I'm All In $29.95, or $60, $199.95 and $299.95 a year. Checkout was in US dollars only and offered card, Alipay and crypto.
Does SpicyChat allow NSFW chat?
The site is marketed for adults, and adult cards appear after you confirm you are over 18 in your profile. Both of our adult test chats ended with the characters slowing things down in character, so we did not confirm how explicit the free model gets.
How good is SpicyChat's memory on the free plan?
Weak in our test. The character recalled 1 of 4 planted facts 13 to 16 messages later and invented the other three; longer memory is sold in the paid tiers.
How do I cancel a SpicyChat subscription?
Card subscriptions are cancelled from your profile under Manage Subscription. Crypto and Alipay payments are one-time and do not renew.