Why AI Benchmarks Are Dead (and What to Track Instead)
TBPN's John Coogan and Jordi Hays explain why public AI benchmarks are broken and how teams evaluate models from Anthropic, Meta, and Google.
Model Mayhem, GPT-6 Astra, Why Nvidia Bought Hugging Face | Diet TBPN
With John Coogan, Jordi Hays, Tyler · Sunday, September 6, 2026
John Coogan and Jordi Hays break down an unprecedented wave of AI model releases, including Anthropic's Claude Fable 5.1, Google's Gemini 3.8 Flash, and Meta's Mu Spark 1.3, culminating in OpenAI's launch of GPT-6 Astra. They analyze Nvidia's $13 billion acquisition of Hugging Face, enterprise AI revenue concentration data from Ramp, and the pivot away from consumer AI companion bots toward enterprise software.
Why 1% of Companies Drive 80% of Enterprise AI Spend
Ramp data shows 1% of companies drive 80% of OpenAI and Anthropic revenue. Here is how consumption pricing changes enterprise software economics.
Why Nvidia Bought Hugging Face for $13B
Nvidia acquired Hugging Face for $12.93B to capture open-source developer mindshare. Here is what builders can learn from their rise.
GPT-6 Astra Hits 99.9% on Arc AGI 3
OpenAI launched GPT-6 Astra with a 99.9% score on Arc AGI 3, pushing benchmark creators to focus Arc AGI 4 on open-ended invention.
Super Grok Ends AI Companions as Frontier Labs Pivot to Enterprise
xAI ends Super Grok romantic companions as frontier labs abandon consumer hype to capture enterprise software revenue.