Breaking news straight out of the AI benchmarking world, and this one actually matters.
Artificial Analysis, the independent benchmark everyone trusts to grade these models fairly, just dropped a report comparing Google's Gemini 4 Argon (high) against OpenAI's GPT-6 Astra (max).
On the Intelligence Index, the two models are now neck and neck. No clear winner there.
But the real headline is hallucinations, the thing nobody wants from an AI model.
Hallucination means the model confidently gives you an answer that's just plain wrong, like someone bluffing through a job interview question.
Gemini 4 Argon clocked in at a 15% hallucination rate.
GPT-6 Astra? A staggering 51%.
That's not a small gap. That's roughly a 70% reduction in hallucinations for Gemini, while matching Astra on raw intelligence.
For anyone building real products on top of these models, that difference is the line between 'trustworthy assistant' and 'needs fact-checking every single time.'
So the real question isn't who's smarter anymore. It's who you'd actually trust with your next critical decision.
🔗 Original Source & Reference: https://www.techmeme.com/260930/p53#a260930p53
Published automatically via FeedMind AI Content Pipeline.
Top comments (0)