This video documents a comparative test of five AI tools—ChatGPT 5.1, Claude Opus 4.5, Gemini 3, and the smart browsers Atlas and Comet—to determine which is most effective at finding Black Friday deals for a sectional couch. The methodology involved using a simple, context-aware prompt for the browsers and a more detailed, structured prompt for the standalone language models.
The results show a clear hierarchy of performance. ChatGPT 5.1 was the decisive winner, identifying the best-value deals and presenting them clearly, though it failed to provide direct links. The Comet browser offered a good user interface but delivered price-inaccurate and retailer-biased results. The Atlas browser failed completely, returning only generic articles. Both Claude and Gemini proved largely useless for the task, providing poorly categorized results without specific product links. The key takeaway is that for a complex economic search like deal-hunting, a well-prompted ChatGPT 5.1 is superior, while agentic browsers may perform better with simpler, more direct queries.
AI Black Friday Deal-Hunting Showdown
0:00:02 The Test: Five AIs (ChatGPT 5.1, Claude Opus 4.5, Gemini 3, Atlas, Comet) are tested to find the best Black Friday deal on a sectional couch. Browsers were given simple prompts, while LLMs received a detailed prompt.
0:00:48 Key Takeaway - Intent is critical: The more specific your prompt (e.g., color, size, price), the better the AI's output. Vague prompts lead to irrelevant results.
0:02:19 Atlas Browser (Failure): This browser completely failed the task. It returned only general articles about deals instead of finding specific, actionable product listings.
0:03:08 Comet Browser (Grade: C): Comet provided a good user interface with specific products but was not very useful. Its results were heavily biased towards Walmart, ignored unspecified details like color, and were not price-optimized, suggesting products well outside the implicit price range.
0:04:31 ChatGPT 5.1 (Winner, Grade: B): Using a detailed prompt, ChatGPT found the highest-value deals, including several couches under $500. It organized results in a clean table with discount percentages. Its main flaw was failing to provide clickable links. It was the best "deal hunter."
0:05:46 Claude Opus 4.5 (Grade: C+): Claude's output was a mess. Its price categories were nonsensical, and it only provided links to retailer homepages instead of specific product pages, which is not helpful.
0:07:20 Gemini 3 (Grade: C-): Gemini performed even worse than Claude. It offered fewer options, unclear categorization, and provided no links whatsoever.
0:09:18 Final Ranking & Conclusion: For this economic search task, ChatGPT 5.1 is the most effective tool. Comet is a distant second, valued more for its interface than its accuracy. Claude and Gemini are not suitable for this task. The experiment suggests that different AIs are good at different things, and for deal-hunting, a properly prompted ChatGPT is currently king.