OpenAI has launched two new AI models, GPT-6.2 Astra and GPT-6.2 Sol, hoping to regain the top spot from Anthropic. Astra scores well on independent tests and is much cheaper to run, but many early users say it is uneven, hard to follow, and disappointing for coding work.
Anthropic’s Claude Fable 5.2 still leads the Artificial Analysis Intelligence Index, while OpenAI’s Astra is second. Astra does well in science, legal work, robotics, and 3D projects, but testers report that it can stop work too early, miss key details, and produce confusing answers.
Anthropic is preparing for an IPO that could raise more than $90 billion, adding to its existing cash reserves. That money could fund more computing power, research, talent, and product development — resources that may help it widen its lead rather than simply defend it.
For ordinary users, this means AI tools will keep improving, but no single company is safely ahead. OpenAI still has strong models and lower costs, while Anthropic has strong reviews and fresh funding; the winner will be decided by real-world usefulness, reliability, and price.
IN AI, THE LEADER IS ONLY AS SAFE AS ITS NEXT MODEL.
Sanjay Sahay
Have a nice evening.

