Free-Access Census: Corrections to the August 2026 Findings

“Free” can mean an ongoing hosted plan, a renewable but infrequent diagnostic, a grant of credits that never replenishes, a timed trial of a paid tier, or software you host with your own model keys. Those offers answer different buying questions. A checkmark for “has a free tier” does not tell you which one you are getting.

Editorial correction — 5 September 2026. The August 19 version claimed a complete partition of 76 software tools, including 16 ongoing free tiers and 36 published-length trials across overlapping sections. That classification is withdrawn, not remeasured against a larger index. It mixed eligibility with marketing labels, counted a free SEO plan without establishing its AI-monitoring allowance, and treated unclear credit renewal as indefinite use. The historical trial registry was a live-field query rather than a pinned record of that cohort; available daily snapshots do not identify the exact original classification run. The old modal-length totals and “nothing free” roster are no longer verified findings. This correction preserves the useful distinctions and evaluation workflow without claiming a new exhaustive census or testing any signup.

Start with what continues after the first visit

Offer	What makes it different	What to establish before relying on it
Ongoing hosted free plan	Access remains available without a subscription	Which feature is included, how allowance renews, cadence, retention and caps
Periodic free diagnostic	A new audit becomes available on a stated cycle	Reset period and whether it measures real answers or technical readiness
Signup credits	A balance is granted once	Whether it expires or replenishes, action costs and paid top-ups
Timed trial	Access ends or converts after a deadline	Eligible tier, duration, card requirement, cancellation and post-trial charge
One-off free check	A single report or limited set of audits	Whether results are cached and whether reruns are possible
Self-hosted or BYOK software	Software access and operating costs are separate	Licence, installation, provider charges, rate limits and support

These are offer types, not mutually exclusive vendor categories. A tool may have a free plan and a trial of a paid tier, or free diagnostics beside a card-required trial. Count the offer you intend to use rather than assigning one permanent label to the company.

Signup credits: Sleepwalker is not an ongoing free monitoring allowance

Sleepwalker’s homepage pricing section states “Sign up and get 10 free credits”, “No credit card required” and “After that, top up credits”. It lists a €0.02 credit price and €5 minimum top-up, with no subscriptions or seats. Read on 5 September 2026, that is clear evidence of an initial grant followed by pay-as-you-go use. It is not evidence that the ten credits come back each month.

Reading previously stored results for free is a separate benefit from running new checks for free. The same distinction matters when a product lets you keep a dashboard after a trial: retained access can be useful without including fresh monitoring. Sleepwalker’s current homepage says checks start from one credit; do not turn the grant into a fixed number of complete multi-engine runs without the applicable model and action prices.

A standing free price can still leave the reset unanswered

AthenaHQ’s Essential plan is labelled “Free”, with “$25 free credit” and “Includes 300 credits”. That establishes a free entry offer. It does not establish that those credits renew monthly or provide indefinite monitoring. The same page explicitly puts “/month” beside Starter’s credit allowance, making it especially important not to copy that renewal term onto Essential.

The earlier article called the Essential credit standing and contrasted it with other vendors’ one-off grants. That stronger claim is withdrawn. Before a trial or free plan becomes part of a recurring workflow, obtain the reset and expiry rule for that exact tier. A dollar value beside “free credit” is the value of an allowance, not a subscription price or a recurring entitlement.

The plan comparison is also the place to check your required integration: AthenaHQ states that Starter API access is a paid add-on while Enterprise includes it. Being able to inspect a product for free does not establish free programmatic access to its data.

A long reset is not the same as no reset

VisibAI’s Free pricing card states €0, no card, one audit per 90 days, ChatGPT only and 20 queries, with a visibility score and the top three issues. Those are the stated terms read on 5 September 2026, not results of an account test.

The article previously grouped this alongside one-off checks. That obscured a real eligibility difference: a recurring quarterly allowance is ongoing, even if it is too infrequent for a daily monitoring workflow. Its suitability depends on what you need to observe. For a periodic diagnostic it may be useful; for a campaign needing repeated comparable checks it may not provide enough observations. Do not call it a one-time offer simply because you prefer a faster cadence.

The vendor also lists a one-off Starter audit with no subscription and a monthly Pro plan. Those are separate paid paths, not evidence that the Free audit includes their engines, exports or rerun frequency. Read the Free column on its own.

A trial belongs to a named plan

Prizvox’s pricing page, read on 5 September 2026, labels the Audit offer “Start 14-day trial” and says “Card required. Cancel before day 14 and you won’t be charged.” Advisor and Agency instead say “Charged today.” The old description that its paid plans generally carry that trial was too broad. Eligibility attaches to Audit, not automatically to every paid tier.

The earlier article also described three free audits without a card. This correction has not established that offer from the pricing page now read, so it is not retained as an available alternative. That is not a claim that Prizvox has no other free offer: it is a limit on the evidence checked here. Its cancellation FAQ says data remains accessible for 30 days after cancellation and advises exporting audit history before then, which is a separate deadline to put in an evaluation plan.

Scrunch’s Core pricing card advertises a seven-day free trial. It illustrates a shorter stated window, not a market-wide modal length. Prepare your prompts, reviewers and export requirements before starting any trial; a longer advertised trial does not necessarily include the feature you need to evaluate.

A “Try for free” button by itself does not establish duration, card requirements, eligible regions, included capabilities or automatic conversion. Likewise, a duration reported in launch coverage is not the vendor’s current contract. This is why the previous HubSpot AEO 28-day entry is not retained as a confirmed vendor-published trial length. Confirm terms on the signup path before scheduling an evaluation, without assuming that an old comparison table is a promise.

Self-hosted and BYOK: preserve the distinction between software and operations

The original article usefully pointed readers to auto-geo, Archytas AISpy, Canonry, Syntropic AI Visibility Audit, GEOrank and geoSurge as alternatives to an ordinary hosted subscription. Keep these as investigation starting points, not a newly verified roster of unlimited or zero-cost monitoring. Their full current licences, deployment paths and allocations were not re-verified in this correction.

For each, check the actual distribution channel and licence rather than treating “open-source”, “source-available”, “desktop download” and “free hosted plan” as interchangeable. Determine whether the tool includes a dashboard, scheduling and storage, or expects your team to provide them. The old blanket statement that none ships a dashboard was not defensible alongside its own description of Canonry’s dashboard and is withdrawn.

A bring-your-own-key setup can still generate provider charges. Running it yourself also means handling installation, upgrades, secrets, backups and any API-provider limits. “No software subscription” does not mean no prompt ceiling, no rate limit, or no operating bill. A hosted demo does not establish unlimited production use of the self-hosted software, and a downloadable codebase does not guarantee that a nontechnical buyer can operate it without help.

Hosted free plans: check the AI feature, not the product badge

The earlier shortlist also linked CiteLens, RankLens, Promptwatch, Ubersuggest, Amplitude AI Visibility, SEORCE, AI Sightline, EdenRank and CheckThat.ai. Their historical numerical allocations are not reissued here as current entitlements. The links remain useful places to begin a plan-specific check.

In particular, an SEO toolkit’s free plan does not prove that its AI-answer monitoring is included. Check the AI-specific prompt allowance, engine list and refresh cadence. A free beta can be temporary even when another tier is called Free. An API key can exist without including the monitoring endpoint or enough calls for your workflow. A plan with broad engine coverage can still omit the particular engine your buyer uses.

An undisclosed cap is unknown, not unlimited. A missing price record is also unknown, not proof that a vendor offers nothing free. That distinction is why the former categorical roster of vendors with “nothing free at all” is withdrawn rather than silently updated from today’s listings. A recorded absence on one pricing page cannot establish an absence across every product, promotion or sales-assisted evaluation.

Run an evaluation that survives the offer ending

Before starting, write down the decision you need to make: whether answers are inspectable, whether the correct markets and engines are available, whether citations link to their source pages, and whether the required data can be exported. Use the same prompt panel and observation conditions across candidates where their products permit it. Record missing observations rather than treating them as zero visibility.

Then save the offer terms: exact plan, free or trial allowance, reset or expiry, card requirement, first charge and cancellation deadline. Ask whether changing prompts or engines consumes allowance, whether a cached result is a new observation, and whether manual reruns are possible. Check whether exports, API access and retention belong to the evaluated tier or a higher one.

A small free allocation can establish whether the interface and output fit your workflow. It cannot by itself establish a reliable market share, a causal improvement in visibility, or the cost of a much larger deployment. Do not spend the trial preparing data that could have been prepared before signup, and export the evidence needed for your decision before access ends.

Method and limits

This remains an archived research article with selectively sourced corrections, not a newly verified count or a substitute current census. Its original research genre and historical population declaration are retained; unsupported findings are explicitly withdrawn rather than made admissible by changing the genre. Public pages cited above were read on 5 September 2026; no accounts were created and no trials or billing flows were executed. Other linked listing pages retain their own source and verification dates. The original publication and verification clocks are preserved rather than relabelled as a fresh census.

For the correction record, the August 19 article’s original partition was 16 ongoing-free tools, 33 trial-only tools with a published duration, 3 trials of unstated duration, 5 one-off-only offers and 19 with nothing free. Its overlapping trial table reported 36 trials, of which 18 ran for a week and 12 for fourteen days. These are the withdrawn assertions being corrected, not retained measurements or current eligibility counts. The roster-conformance query only searched for free or zero-dollar wording; it could not establish recurring credits or AI-feature eligibility. Repeating that old arithmetic here documents what was withdrawn, without supplying the missing historical proof.

See the billing-model walkthrough for the paid commitment and the buyer checklist for the wider product decision. If a comparison gives you only a “free” checkbox, use the offer types above to recover the question it left unanswered.