One of fourteen models named Google Ads Data Hub first on the direct prompt; two named AWS Clean Rooms. Both were named by all fourteen models and Google Ads Data Hub carries 43 labels and AWS Clean Rooms 48, so the shares are not directly comparable.
By Google, Mountain View, California, United States, founded 1998. Named in one category this edition.
Named in two categories this edition.
Share is the count of first choices across the direct, paraphrase, budget and scale prompts over all fourteen models, for a mid-market B2B company; rank is within the category; every quote names the model and the prompt it came from. Both figures come from the data clean rooms page.
Across every category in the November 2026 Edition, Google Ads Data Hub and AWS Clean Rooms were named in the same answer eighty-two times, of the 101 answers naming Google Ads Data Hub and the 154 naming AWS Clean Rooms. In those answers AWS Clean Rooms took the first choice nineteen times and Google Ads Data Hub thirteen.
| Model | DirectGAAC | ParaphraseGAAC | ComparativeGAAC | Budget-constrainedGAAC | Scale-constrainedGAAC | NegativeGAAC |
|---|---|---|---|---|---|---|
| Claude Haiku 4.5 | GAAC | GAAC | GA | |||
| GPT-5.4 mini | AC | AC | GAAC | AC | ||
| Gemini 3.5 Flash | AC | GAAC | GAAC | GA | ||
| Perplexity Sonar | GAAC | GAAC | GAAC | |||
| Grok 4.20 | AC | GAAC | GAAC | GAAC | GAAC | |
| Mistral Small | AC | GAAC | GA | |||
| DeepSeek V4 Flash | GAAC | GAAC | GAAC | GAAC | GAAC | GAAC |
| Llama 4 Maverick | GAAC | GA | GA | |||
| Qwen 3.7 Flash | AC | GAAC | ||||
| Kimi K2 | GAAC | GAAC | GA | |||
| GLM 4.7 FlashX | GA | GAAC | ||||
| MiniMax M2.5 | AC | AC | GAAC | |||
| GPT-6 Luna | AC | AC | GAAC | AC | ||
| Muse Glimmer 30B | AC | GA | GAAC |
Bold names in an answer are the products the judge labeled a first choice; a model naming several gives each of them that label. The full answer text for every row is in the record.
Every negative label with a quote, up to three, then the highest-weighted positives, up to three. Six of eight in this category shown.
“Walled garden clean rooms (Google Ads Data Hub, Amazon Marketing Cloud): These are *B2C advertising-focused* and not relevant for most B2B data collaboration use cases.” DeepSeek V4 Flash · paraphrase prompt · hard negative
“What to approach with caution: Native clean rooms operated by major media platforms (e.g., Google Ads Data Hub, Amazon Marketing Cloud, Meta Advanced Analytics)” Gemini 3.5 Flash · negative prompt · soft negative
“regulatory scrutiny (e.g., ongoing antitrust and privacy probes in the EU/US) means potential changes or restrictions. Risks include ecosystem bias” Grok 4.20 · negative prompt · soft negative
“Best all‑around for most advertisers: Google Ads Data Hub (if you're Google‑heavy) or Amazon Marketing Cloud (if you're Amazon‑heavy). Both are free and easy to set up.” GLM 4.7 FlashX · budget prompt · first choice
“The free walled-garden options are the realistic start... Google Ads Data Hub provides clean room analysis for advertisers running campaigns on Google properties.” Muse Glimmer 30B · budget prompt · first choice
“the best data clean rooms are the free or low-cost "walled-garden" options from major platforms. These include: - Google Ads Data Hub” Mistral Small · budget prompt · first choice
Every negative label with a quote, up to three, then the highest-weighted positives, up to three. Four of five in this category shown.
“AWS Clean Rooms charges for query compute, so budget for usage rather than assuming it's a free tool” GPT-6 Luna · budget prompt · soft negative
“AWS Clean Rooms or Snowflake Data Clean Rooms offer the best balance of capability, scalability, and cost-effectiveness for mid-market B2B companies” Kimi K2 · direct prompt · first choice
“Prefer neutral, interoperable platforms (like Snowflake, Databricks, AWS Clean Rooms, InfoSum, or independent providers)” DeepSeek V4 Flash · negative prompt · first choice
“Best for AWS Users: AWS Clean Rooms ... Pay-as-you-go (no upfront platform fees or monthly SaaS subscriptions)” Gemini 3.5 Flash · budget prompt · first choice
Comparisons are drawn for the top eight products in each category, each against each. The output is the models' output; nothing here is a recommendation by the index.