method_version 2026.10 · reproducible

The Method

Seven weighted dimensions over public, checkable fields. Every value names its official source. A field that is not verified is never scored as zero or as a midpoint. bestaicharacterchat.com checks the public policy, pricing and data-control pages of AI character chat platforms against one published method, and publishes a platform's order only after every scored field is verified on its official pages.

Pool and exclusions

A platform enters the pool when its core product is a conversation with a chosen or created character. Group rooms are allowed. Single-persona companion apps, roleplay campaign engines and general-purpose assistants are out of scope; other publications in the network cover them. One platform was removed during scoping under a child-safety rule and is not evaluated or compared on this site.

In this method, a free tier is a no-payment plan that the official page itself describes. A vendor's word "free" on its own is recorded as vendor-stated, and does not count as a free-tier value.

Seven dimensions

Dimension	Weight	Scoring rule (official field)
D1 Free tier shape	15%	Unlimited text stated on the official site = 100; daily quota = 70; limited trial = 40; no free tier = 0.
D2 Price transparency	15%	Every paid tier priced on the official site = 100; starting price only = 50; no public price = 0.
D3 Sign-up requirement	10%	No sign-up needed = 100; email or account = 60; card required = 0.
D4 Content-policy disclosure	15%	Content-policy page on the official site = 50; age threshold stated = 50. This scores whether rules are published, not how strict they are.
D5 Data and account control	20%	Chat export route 25; account deletion route 25; training opt-out 25; public privacy policy 25.
D6 Context disclosure	15%	Context window published in tokens = 100; description only = 50; none = 0.
D7 Platform coverage	10%	Web, iOS and Android each listed on the official site = 1 point; points divided by 3, then multiplied by 100.

The weights total 100%: 15 + 15 + 10 + 15 + 20 + 15 + 10.

Scoring formula

free_score      = {unlimited: 100, daily_quota: 70, limited_trial: 40, none: 0}
price_score     = {per_tier_public: 100, from_price_only: 50, none: 0}
signup_score    = {none_required: 100, email_or_account: 60, card_required: 0}
policy_score    = 50 * content_policy_page + 50 * age_gate_stated
control_score   = 25 * export + 25 * account_delete + 25 * training_opt_out + 25 * privacy_policy_public
context_score   = {tokens_public: 100, description_only: 50, none: 0}
platform_score  = (web + ios + android) / 3 * 100

total_score = 0.15*free + 0.15*price + 0.10*signup + 0.15*policy
            + 0.20*control + 0.15*context + 0.10*platform

The formula is published so that any score can be recomputed from the field values. No platform has been scored in this edition, so no total is shown, and the shortlist is not ordered by any total.

Unverified fields

A field the official page does not state is recorded as not yet verified.
An unverified field is never scored as 0 or 50.
A platform is scored only when every scored field is verified. Until then it stays on the shortlist with no order.
A page that mentions a mechanism without the full field, for example a deletion request with no steps, counts as partial. Partial is not verified.
A vendor's marketing line, such as "free", is recorded as vendor-stated. It does not satisfy a field that needs a classified value.
"Not yet verified" means this edition did not confirm the field from an official page. It does not mean the platform lacks the term, and it is never read as a negative answer.

Verification matrix, 2026.10

Legend: yes = value stated on the official page; partial = mechanism mentioned but not the full field; vendor-stated = marketing claim without the classified value; — = not yet verified. Verified cells in this matrix: 7 of 80.

Platform	Free tier	Price	Sign-up	Content rule	Age statement	Export	Deletion	Context window
Chai	—	—	—	—	—	—	—	—
Character.AI	yes	yes	—	—	yes (see conflicts)	partial	partial	—
Chub AI	—	—	—	—	—	—	—	—
CrushOn.AI	—	—	—	—	—	—	—	—
DreamGen	—	yes	yes	—	—	—	—	yes
Janitor AI	—	—	—	—	—	—	—	—
PolyBuzz	vendor-stated	—	—	yes	—	—	partial	—
SillyTavern	—	—	—	—	—	—	—	—
SpicyChat	—	—	—	—	—	—	—	—
Talkie AI	—	—	—	—	—	—	—	—

Sources and independence

Official pages are the only source of field values. Third-party pages are listed so that readers can see what was checked and what was not used. T3 shares a source with bestuncensoredaichat.com, a sibling site in this network, as the table notes.

ID	Source	Captured	Independence	Use on this site
O1	Character.AI subscribe page, monthly and annual views	16 Oct 2026	Official, primary	Prices, free-tier table, renewal and cancellation wording
O2	Character.AI privacy policy, effective 1 July 2026	16 Oct 2026 (browser)	Official, primary	Age statement, training use, data categories, retention wording, rights list, account-deletion route (partial)
O3	DreamGen pricing page	16 Oct 2026	Official, primary	Plans, list prices, context windows, sign-up link
O4	PolyBuzz homepage and FAQ	16 Oct 2026	Official; vendor self-description marked as such	Content-display rule, free-access statement, footer links
O5	SillyTavern homepage	16 Oct 2026	Official, primary	Self-description and links only
T1	Third-party review dated 2 Oct 2026	16 Oct 2026 (summary only)	Third party; research-based summary, read as text	Cross-check only; its figures are superseded by official pages
T2	Third-party comparison of companion apps	16 Oct 2026	Third party; relationship to listed platforms not checked	Method reference only
T3	Third-party free-tier roundup	16 Oct 2026	Third party; shares its source with bestuncensoredaichat.com (Uncensored Chat Bench), a sibling site in this network	Category approach only; no facts cited
T4	Third-party roundup of roleplay platforms	16 Oct 2026	Third party; relationship to listed platforms not checked	Category approach only

Known conflicts

Topic	What the sources say	Resolution
Character.AI age statement	A third-party review and a vendor-run roundup state opposite positions on minors. The official privacy page, effective 1 July 2026, reads: "The Services are not designed for minors under 13, and our Terms of Service prohibit use by anyone under 13 or under 16 if you are a resident of the European Economic Area or the United Kingdom."	The official current policy governs, and this site takes no position between the two roundups. Read the official privacy page before you sign up.
DreamGen content rule	One third-party article describes the platform as both allowing and restricting such content.	Not stated on this site. The official content-rule text was not captured.
DreamGen price	A third-party roundup lists $8 per month.	The official pricing page shows $7.83 for Starter, and $6.26 under a 20% first-month offer. Official figures are used.
Character.AI monthly price	A third-party review (T1) reports $9.99 per month.	The official subscribe page shows $9.99 on the monthly view. Official figure used; the two agree.
Character.AI free tier	A third-party review reports ads and queueing in the free tier.	The official table shows full ads and "Slow mode: Yes" in the free column. Only the official wording is used.

Limitations

Each capture is a single read. Pages change without notice.
Several official pages could not be captured in this edition, so their platforms show no verified fields.
No account was created, no platform was used, and no output was evaluated.
Vendor self-descriptions are recorded, but never used as verified values.
Nothing on this site is legal, medical or financial advice.

Edition log

2026.10, first edition: 10 platforms on the shortlist, 7 of 80 scored fields verified, no platform scored, no order published.

Disclosure

bestaicharacterchat.com is part of a network of independent review publications. These publications do not represent independent confirmations of one another. The full list of related publications is on the about page.