Ziina disappeared from the payment-link answer
The GPT-5.5 control named Ziina. All 3 GPT-5.6 variants omitted it. Sol and Luna dropped the brand, while Terra named no brands at all on launch day.
Home / GPT-5.6 brand visibility
Release field note
What changed for brands on GPT-5.6
The same buyer questions produced a materially different set of winners and omissions on launch day. A second capture round showed which changes persisted and which rankings were still moving.
4 of 6 shortlists changed
3 GPT-5.6 variants compared
Week 2 resample
Launch day
A model release changed 4 of 6 brand shortlists
The comparison used the same six queries, the same model-direct surface and captures taken on the same day. Against the GPT-5.5 control, four GPT-5.6 shortlists changed.
For a brand owner, the important point is not that every answer changed. It is that a release altered who made the shortlist, who led it and how concentrated the recommendation set became.
Cal.com took #1 from SavvyCal
SavvyCal led the GPT-5.5 control. Cal.com ranked #1 in Sol, Terra and Luna, making the change unanimous across the GPT-5.6 variants.
The BNPL list went from 4 names to 2
The control listed Tabby, Tamara, Postpay and Cashew. Each GPT-5.6 variant kept only Tabby and Tamara, halving the shortlist.
The variants picked different UAE gateways
Sol picked Stripe, Terra picked Telr and Luna picked PayTabs. The model family did not produce one shared answer to the same gateway question.
Week 2
The large signals held. The edges drifted.
The identical six queries were captured again 10 days later across the same GPT-5.5 control and GPT-5.6 Sol, Terra and Luna lanes. The durable findings were the ones repeated across dates and lanes.
Held fully
Ziina stayed out of every GPT-5.6 payment-link answer
The GPT-5.5 control still named Ziina. Sol, Terra and Luna still did not. Terra now named 4 competing providers and continued to omit Ziina.
Strengthened
Cal.com held #1 and the control joined it
Cal.com remained #1 on every GPT-5.6 variant. The GPT-5.5 control also flipped to Cal.com, leaving SavvyCal at #2 in every lane.
Mostly held
Tabby and Tamara stayed the BNPL default
Tabby stayed #1 and Tamara #2 in every lane across both dates. Sol and Terra still named only that pair. Luna drifted by adding Postpay as a hedged third option.
Partly converged
The gateway leaders narrowed from 3 to 2
Sol moved from Stripe to PayTabs, Terra held Telr and Luna held PayTabs. The three different launch-day leaders became 2 distinct leaders, but that rests on one single-cell move.
How to read it
Treat repeatability as the signal
The GPT-5.5 control also drifted on 3 of 6 queries. SavvyCal gave way to Cal.com, Ziina dropped from the gateway query, and Cashew was replaced by Spotii on BNPL. The observed drift was not unique to GPT-5.6.
The honest caveat: n=1 per cell per date
Each lane was sampled once on each date. A launch-to-week-2 difference could be sampling variance or a serving-stack change, and this dataset cannot separate the two. Single-cell moves such as Sol changing its gateway pick and Luna adding Postpay should be treated as possible noise. A third capture round would be needed to call them genuine shifts.
Watch omissions
Ziina's repeated absence across every GPT-5.6 variant is stronger evidence than one rank change in one lane.
Separate leaders from tails
Cal.com and Tabby held the leading positions while lower-ranked names continued to move.
Resample before reacting
The repeat capture distinguished the durable shortlist changes from the observations that still look noisy.
Your category
See which brands the answer engines recommend
Explore the automated WhoCitesYou audit for your brand.
Explore the audit