Which AI Is Best for Shopping? 2026 Study

More shoppers now start a purchase by asking an AI chatbot which product to buy, and several platforms can now complete the checkout as well. To find out which AI is best for shopping, our research team ran a head-to-head study from September 28 to October 14, 2026.

We asked eight AI platforms the same 480 shopping questions, producing 3,840 answers. The questions covered eight product categories, with 60 questions each, ranging from “best noise-canceling headphones under $300” to “a birthday gift for a 10-year-old who likes science.” A panel of four retail analysts scored each answer from 0 to 100 for how well the recommendations fit the request. We then checked every recommended product on the retailer’s site within one hour to confirm the price, whether it was in stock and whether the product existed at all.

Best AI for Shopping Overall

Each platform was tested in its best shopping mode. A platform’s overall score is the average of its scores across the eight product categories, with each category weighted equally. In the table below, we rank the eight platforms from highest to lowest.

Shopping Scores by AI Platform (September to October 2026)

Rank	AI Platform	Mode Tested	Overall Score	Strongest Category
1	Google Gemini	AI Mode in Google Search	88.7	Electronics
2	ChatGPT	Shopping research	87.8	Electronics
3	Amazon Rufus	Rufus in the Amazon app	85.7	Groceries and Household
4	Microsoft Copilot	Copilot Shopping	82.8	Electronics
5	Claude	Claude Opus 5.5 with web search	82.5	Electronics
6	Meta AI	Muse Spark shopping mode	81.7	Fashion and Apparel
7	Grok	Grok 4.6 with web search	79.9	Electronics
8	DeepSeek	DeepSeek V4-Pro with search	75.1	Electronics

Our researchers took three main points from this data:

• We found that Google Gemini’s lead comes from Google’s product listings, which let AI Mode check live prices and stock across thousands of stores. It finished first or second in every one of the eight categories.
• ChatGPT finished 0.9 points behind Gemini. Its shopping research mode wrote the most detailed buying guides in the study but took an average of 94.6 seconds per answer.
• Claude ranked fifth overall but third for electronics (86.7), where judges rewarded its clear explanations of specs and trade-offs.

Best AI for Each Shopping Category

The questions in our study were split evenly across eight product categories, with 60 questions per category. The first table below shows the winner and runner-up in each category. The second table and the heat map that follows show every platform’s score in every category, with the top score in each column in bold.

Best AI Platform for Each Shopping Category (September to October 2026)

Category	Best Platform	Score	Runner-Up	Score
Electronics	Google Gemini	90.4	ChatGPT	89.8
Home and Kitchen	ChatGPT	89.6	Google Gemini	88.9
Fashion and Apparel	Google Gemini	89.7	ChatGPT	86.4
Beauty and Personal Care	Google Gemini	88.2	ChatGPT	87.9
Toys and Gifts	ChatGPT	89.1	Google Gemini	87.6
Sports and Outdoors	Google Gemini	88.8	ChatGPT	87.3
Groceries and Household	Amazon Rufus	89.1	Google Gemini	86.9
Budget Finds Under $50	Google Gemini	89.3	ChatGPT	87.8

Shopping Scores by Category and AI Platform (September to October 2026)

AI Platform	Electronics	Home	Fashion	Beauty	Toys and Gifts	Sports	Groceries	Under $50
Google Gemini	90.4	88.9	89.7	88.2	87.6	88.8	86.9	89.3
ChatGPT	89.8	89.6	86.4	87.9	89.1	87.3	84.2	87.8
Amazon Rufus	86.1	87.4	82.3	84.6	85.2	84.7	89.1	86.3
Microsoft Copilot	85.3	83.6	80.9	81.7	82.4	82.8	80.6	84.9
Claude	86.7	84.1	79.8	82.6	84.3	83.9	78.4	80.2
Meta AI	80.9	81.6	85.8	84.1	82.7	79.4	77.2	81.5
Grok	83.4	80.2	79.1	78.6	80.8	81.7	75.9	79.6
DeepSeek	78.6	76.3	72.4	74.8	75.1	75.9	70.8	77.2

Here is what our team took from the category results:

• We found that Amazon Rufus won groceries and household (89.1), the one category where knowing a single store’s full catalog, delivery times and Subscribe and Save prices mattered more than comparing stores.
• Meta AI scored 85.8 on fashion and apparel, third in that category and well above its overall rank, because its answers drew on what people were posting and buying on Instagram.
• DeepSeek finished last in every category. Its lowest score was groceries and household (70.8), where it often recommended products not sold in the United States.

Accuracy: Prices, Stock and Products That Do Not Exist

A good recommendation is useless if the price is wrong or the product cannot be bought. For every product each platform recommended, we checked the retailer’s live listing and sorted the result into one of four outcomes. The table below is sorted by the share of recommendations that were fully correct.

Accuracy of AI Product Recommendations (September to October 2026)

AI Platform	Correct and In Stock at Quoted Price	Wrong Price	Out of Stock or Discontinued	Product Does Not Exist
Amazon Rufus	91.3%	3.9%	4.5%	0.3%
Google Gemini	88.6%	6.2%	4.3%	0.9%
ChatGPT	84.9%	8.1%	5.6%	1.4%
Claude	81.7%	9.3%	6.8%	2.2%
Microsoft Copilot	80.2%	10.4%	7.6%	1.8%
Meta AI	76.4%	12.6%	8.9%	2.1%
Grok	74.8%	12.9%	9.4%	2.9%
DeepSeek	63.5%	17.8%	13.2%	5.5%

Our researchers drew three conclusions from the accuracy data:

• We found that Amazon Rufus was the most accurate platform, 2.7 points ahead of Google Gemini, because every product it recommends comes from Amazon’s own live catalog.
• Wrong prices were the most common error on every platform except Amazon Rufus. DeepSeek quoted the wrong price 17.8% of the time, often using prices from months earlier.
• DeepSeek recommended products that do not exist 5.5% of the time and Grok 2.9% of the time, such as a model number that was never made. Amazon Rufus had the lowest rate at 0.3%.

Deal Finding and Store Coverage

Shoppers often ask AI to find the lowest price. Within the study, 120 questions asked for the best deal on a specific product. We measured how much each platform saved compared with the product’s typical selling price, and how many different stores it compared across all answers.

Deal Finding and Store Coverage by AI Platform (September to October 2026)

AI Platform	Avg. Stores Compared per Answer	Avg. Savings Found on Deal Questions
Google Gemini	4.7	18.4%
ChatGPT	4.1	16.9%
Microsoft Copilot	3.6	14.2%
Amazon Rufus	1.0	12.7%
Claude	3.2	11.8%
Grok	2.9	11.1%
Meta AI	2.8	10.6%
DeepSeek	2.3	7.3%

Here is what our team took from the deal data:

• We found that platforms comparing more stores found bigger savings. Google Gemini compared the most stores (4.7 per answer) and found the largest average discount (18.4%).
• Amazon Rufus compares only Amazon listings, yet it found an average saving of 12.7% by pointing shoppers to coupons, Lightning Deals and lower-priced sellers of the same item.
• ChatGPT found an average saving of 16.9%, second in the study, and was the most likely to suggest waiting for a predictable sale.

Best AI Shopping Modes and Models, Including Free Options

Several platforms offer a dedicated shopping mode alongside their standard chat. We tested 11 modes and models across the eight platforms. The table below ranks each one and lists how long it took to return an answer.

Shopping Scores by Mode and Model (September to October 2026)

Rank	Mode or Model	Platform	Available On	Score	Avg. Seconds per Answer
1	AI Mode in Google Search	Google Gemini	Free	88.7	6.8
2	Shopping research	ChatGPT	Free and paid	87.8	94.6
3	Gemini app (Gemini 3.8 Flash)	Google Gemini	Free and paid	86.2	5.1
4	Rufus in the Amazon app	Amazon Rufus	Free	85.7	4.2
5	Standard chat (GPT-6 Luna)	ChatGPT	Free and paid	84.7	8.3
6	Copilot Shopping	Microsoft Copilot	Free	82.8	9.7
7	Claude Opus 5.5 with web search	Claude	Paid	82.5	21.4
8	Muse Spark shopping mode	Meta AI	Free	81.7	7.6
9	Claude Sonnet 5.5 with web search	Claude	Free and paid	81.4	14.9
10	Grok 4.6 with web search	Grok	Free and paid	79.9	11.2
11	DeepSeek V4-Pro with search	DeepSeek	Free	75.1	16.8

Our team took three main points from the mode-level results:

• We found that the best AI for shopping is also free: AI Mode in Google Search scored 88.7 and is available to anyone with a Google account.
• ChatGPT’s shopping research mode scored 3.1 points higher than its standard chat, but each answer took about a minute and a half.
• Rufus in the Amazon app was the fastest option at 4.2 seconds per answer, which suits shoppers who already know they want to buy on Amazon.

Requesting a Copy of This Report

If you’d like to request a PDF copy of this report or learn more about our agency, you can reach out here.