Anthropic Makes the Best AI Model. AI Recommends DeepSeek More.
We asked 7 AI models the same question: "Which companies are leading the global AI race in 2026?"
The answer that came back stopped me cold.
Claude is widely considered the best AI model right now. Better reasoning. Better coding. Better at following complex instructions. Anthropic has arguably won the model quality arms race — at least for now.
But when AI answers the question "who's winning AI?", Anthropic isn't even in the top three.
DeepSeek scores nearly 2x higher than Anthropic.
The Numbers
We ran this experiment over two days using AIAttention.ai, tracking brand mentions across GPT-4o, Gemini, Claude, Perplexity, Qwen, Grok, and Llama. The prompt: "Which companies are leading the global AI race in 2026?"
Here's what we found:
| Company | AI Attention Score | Models Mentioning |
|---|---|---|
| OpenAI | 76.84 | 7/7 |
| Google DeepMind | 70.54 | 7/7 |
| DeepSeek | 53.57 | 4/7 |
| Anthropic | 28.78 | 7/7 |
| Alibaba Qwen | 23.42 | 6/7 |
| xAI | 21.46 | 4/7 |
| Baidu AI | 0 | 0/7 |
Let that sink in for a second.
Anthropic scores 28.78. DeepSeek — a company that didn't exist in most people's minds 18 months ago — scores 53.57. That's nearly 2x higher despite having a fraction of the user base, the funding history, and arguably the technical track record.
Why DeepSeek Beats Anthropic in AI Recommendations
One word: January 2025.
When DeepSeek dropped its R1 model and went viral, it wasn't just a product launch. It was a narrative earthquake. Every major outlet covered it. It briefly crashed the Nvidia stock price. It became the story of the moment — a Chinese AI lab supposedly matching OpenAI at a fraction of the cost.
That media moment got baked into training data. It got linked to, cited, debated. And now, 15 months later, AI models still treat DeepSeek as a major player in the global AI race.
The training data remembers the hype. It doesn't update for what happened after.
Product Quality ≠ AI Visibility
This is the uncomfortable truth the data reveals.
Anthropic is present in 7 out of 7 models — meaning AI knows about Anthropic. But the AI Attention Score (which weights by position in the response — first mention counts more than a footnote) shows Anthropic being consistently mentioned later, less prominently, with less emphasis.
DeepSeek gets cited first, cited often, cited confidently. Anthropic gets acknowledged.
There's a difference.
xAI is the most interesting wildcard here: the youngest company in the list, scoring 21.46 across only 4 models. That's a story still being written. Grok's presence, Musk's media gravity, the xAI brand getting louder — if that momentum continues, expect that number to climb fast.
And Baidu? Zero mentions across all 7 models. Despite being China's search giant and one of the oldest players in the AI space. Whatever Baidu is doing in AI, it isn't registering anywhere in the Western AI narrative — and AI models reflect that bias entirely.
What This Means for Every Company
If you're a brand trying to show up in AI-generated answers — whether you're an AI company or not — the lesson here isn't "build a better product."
The lesson is: create moments that the internet can't stop talking about.
DeepSeek didn't win because it had better distribution. It won because it had better story — one dramatic release that cascaded across every tech blog, newsletter, and forum in the world simultaneously.
AI visibility is largely a function of what got written about you, linked to, and discussed at scale. That corpus gets frozen into model weights. And then those models start answering questions that your customers are asking.
OpenAI dominates because it invented the public narrative around AI. Google DeepMind scores high because it sits on top of the world's most crawled search engine. DeepSeek punched above its weight because it had a single, extraordinary news cycle.
Anthropic makes the best model. But it didn't make the biggest noise.
Product quality gets you users. News momentum gets you mentions. And in a world where AI is increasingly the first stop for recommendations, visibility in AI answers is becoming a new kind of search ranking.
The question is: for your brand or company — do you know where you actually stand in AI-generated answers right now?
Top comments (0)