Two of fourteen models named Amazon Marketing Cloud first on the direct prompt; one named InfoSum. Amazon Marketing Cloud was named by twelve of the fourteen models and InfoSum by fourteen and Amazon Marketing Cloud carries 24 labels and InfoSum 39, so the shares are not directly comparable.
By Amazon, Seattle, United States, founded 1994. Named in one category this edition.
Named in three 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 October 2026 Edition, Amazon Marketing Cloud and InfoSum were named in the same answer thirty-six times, of the 62 answers naming Amazon Marketing Cloud and the 119 naming InfoSum. In those answers InfoSum took the first choice one time and Amazon Marketing Cloud eight.
| Model | DirectAM | ParaphraseAM | ComparativeAM | Budget-constrainedAM | Scale-constrainedAM | NegativeAM |
|---|---|---|---|---|---|---|
| Claude Haiku 4.5 | AM | |||||
| GPT-5.4 mini | AM | |||||
| Gemini 3.5 Flash | AM | AM | AM | AM | ||
| Perplexity Sonar | ||||||
| Grok 4.1 Fast | AM | |||||
| Mistral Small | ||||||
| DeepSeek V4 Flash | AM | AM | ||||
| Llama 4 Maverick | AM | |||||
| Qwen 3.7 Flash | AM | |||||
| Kimi K2 | AM | |||||
| GLM 4.7 FlashX | AM | AM | ||||
| MiniMax M2.5 | AM | AM | ||||
| GPT-6 Luna | AM | AM | ||||
| Muse Glimmer 30B | AM | AM |
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 DCRs are incredibly powerful, but only within their own ecosystems. They are plagued by severe vendor lock-in.” Gemini 3.5 Flash · negative prompt · soft negative
“Walled Garden Clean Rooms (Google, Meta, Amazon) ... Why to be cautious: Vendor lock-in” DeepSeek V4 Flash · negative prompt · soft negative
“Amazon documents a 100-distinct-user threshold for certain first-party data analyses.” GPT-6 Luna · negative prompt · soft negative
“Amazon Marketing Cloud (AMC) seems particularly relevant if your company has significant Amazon advertising presence... it could be an accessible option for mid-market companies” MiniMax M2.5 · direct prompt · first choice
“set up Google Ads Data Hub or Amazon Marketing Cloud. It will teach your team how to work with clean room query logic without costing you a massive licensing fee.” Gemini 3.5 Flash · scale prompt · first choice
“If ad-focused, begin with free walled-garden DCRs (AMC/Google)—they solve 80% of needs without spend.” Grok 4.1 Fast · budget prompt · first choice
Every negative label with a quote, up to three, then the highest-weighted positives, up to three. Six of eight in this category shown.
“Avoid enterprise SaaS (e.g., LiveRamp, InfoSum) at $500+/month unless trialing.” Grok 4.1 Fast · budget prompt · hard negative
“pricing and complexity are often overkill for mid-market needs” DeepSeek V4 Flash · direct prompt · hard negative
“like LiveRamp (Habu), InfoSum, or Decentriq is rarely feasible, as licensing fees for these platforms often run in the tens or hundreds of thousands of dollars annually” Gemini 3.5 Flash · budget prompt · soft negative
“InfoSum – Praised for its flexible collaboration capabilities and suitability for mid-market needs, without requiring enterprise-scale resources.” Mistral Small · direct prompt · first choice
“I would recommend InfoSum as a privacy-safe data collaboration platform for a mid-sized B2B company.” Llama 4 Maverick · paraphrase prompt · first choice
“I would recommend InfoSum if your priority is privacy-safe data collaboration with partners” Perplexity Sonar · paraphrase 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.