Mocha leaves the index, and we benchmarked it after it had already announced its own end.

A removal about our own record, not just the vendor's calendar

Mocha, published by Axilla, Inc at getmocha.com, no longer holds a row in this index. Most coverage of a closure like this gets written from the outside, by people who never measured the product. That is not our position. We benchmarked Mocha, published its numbers, and kept them live past the point where they still described the product it claims to be. This note is about that failure of our own process, not primarily about the vendor's announcement.

What the vendor has actually said

On 15 May 2026, Mocha's own blog published a post by Nicholas titled "Mocha is shutting down on August 1, 2026," at https://getmocha.com/blog/mocha-shutdown. It gives an economic account of the decision: user acquisition costs made worse by competition, expensive per token AI economics, high support costs, and an inability to raise further funding. The post states that after 1 August, "Mocha's app builder will stop working: no more building, previews, or live apps on our platform," and it points existing users toward Anything, at create.xyz, through a one click migration carrying 20 dollars of credits, plus a manual export tool under Settings then Export. Both the migration link and the export tool are stated to remain available for 30 days after shutdown. The vendor has published nothing since that post, and its blog index still leads with it.

We fetched https://getmocha.com/pricing again on 21 August 2026 before writing this. It carries a site banner reading "Mocha is being sunset. Read more on the blog," alongside the line "Everyone can use Mocha for free and build apps of any size" and, separately, "Paid plan purchases are no longer available." The three paid tiers, Bronze at 20 dollars a month, Silver at 50 and Gold at 200, are all written in the past tense now, in phrases like "for individuals who needed additional features." The free Starter tier with 20 credits is still listed as though nothing had changed.

We want to be precise about what this adds up to, because the vendor's own pages disagree with each other. The stated shutdown date of 1 August 2026 has passed. No plan can currently be bought. The pricing page's claim that anyone can still build for free sits next to a shutdown post that says the builder stops working entirely. We are not writing that Mocha is switched off, and we are not writing that it still works normally, because neither is what the record shows. What is not in dispute, and what matters for this index, is that the product has been formally sunset and that nothing on it is purchasable.

Three things wrong with the row while it was still live

The first is small on its own but worth naming plainly. We published "Free tier with a monthly allowance. Paid plans from EUR 20 per month" and stored a price_from_eur of 20. The real entry price was 20 US dollars a month, not 20 euro, and no currency conversion was applied before we published it. Against a pricing page that no longer sells anything, this error is moot in practice now, but it was wrong on the day we wrote it and would have been wrong regardless of the shutdown.

The second is worse. Mocha's value subscore was 74, the highest of its own ten measurements, ahead of design at 72 and agent performance at 66. The value axis is meant to answer what a reader gets for what they pay. Awarding the single highest score on the entire row to that exact question, for a product whose own pricing page states paid plans can no longer be purchased, does not just understate a problem. It answers the wrong question convincingly, in the wrong direction.

The third is the timing, and it is the one we consider the actual defect in our process rather than a judgement call after the fact. All 45 recorded executions behind Mocha's row are dated between 11 and 15 August 2026, which is after the vendor's own stated shutdown date of 1 August. The published reliability of 53.3 percent and median build time of 105.2 seconds are not a snapshot of a product competing for a reader's business. They describe a service in the process of being switched off, measured as though it were still active in a market it had already announced it was leaving.

Two things we are naming on ourselves, not calling fixes

Two smaller facts belong on this record honestly rather than folded quietly into the removal. The row stayed live and directly reachable at /builders/mocha while our own sitemaps advertised only three Mocha URLs in total: its builder page, its reviews page and its single compare pairing. And the row carried a next scheduled run of 21 August 2026 at 09:00 UTC, a slot that arrived and passed without firing. We were advertising a future measurement of a product that had already announced its own end, and that measurement was never going to happen. Removing the row is what retires that scheduled slot.

Why removal, and not a correction

We could, in principle, fix the price, drop the value score and hold the row with a note attached. We chose not to, because there is no version of this row that actually informs a reader. A reader consults this index to decide what to build on, and the honest answer for Mocha is that the question no longer applies. A corrected row would still imply a decision is available where none is left to make.

If you have a Mocha project and need somewhere to go, the vendor's own recommended path is Anything, at create.xyz. Anything is itself measured in this index, currently rank 20 with a composite of 67.16, and its row carries its own subscores and its own pricing rather than any claim of ours about it beyond what that row already states.

Both of Mocha's product pages that used to lead readers here, and its one compare pairing, now redirect to this note rather than to a page for a product with nothing left to evaluate.