Grok 4.6 Shows How Fast Your AI Options Are Expanding

A year ago, “frontier model” meant a model from OpenAI, Anthropic, or Google. A few months ago, it meant OpenAI or Anthropic. Now, the conversation has to include Chinese open-weight models pushing efficiency and cost, and as of this week, xAI’s Grok is back in the race. Grok 4.6 is posting benchmark numbers that put it in the category of GPT‑5.6 or Fable 5 at a fraction of the cost — though benchmarks are not the real world. After initial testing, users are not ready to declare Grok 4.6 a Fable or GPT class model, but they are ready to argue that Grok and xAI are back in the competition.

Big Money: Cognition, Lovable, and the Neoclouds

Cognition is seeking a new funding round on the back of booming coding agent demand. Bloomberg reports early talks with investors for a valuation of $40 billion, up almost 50% from the $26 billion valuation in its last round just three months ago. Revenue figures back it up: Cognition has doubled its revenue run rate to $1 billion since the last raise, and is seeking $1 billion in this round. The premium attached to coding agents is growing among venture investors. Cursor, a close comp, sought a $50 billion valuation on $2 billion in annualized revenue in March, before xAI acquired it in a $60 billion all-stock deal. If Cognition prices at $40 billion, that deal could start to look like a bargain. Many expect a similar path for Cognition.