Anthropic chief scientist says Claude 4 is coming within six months

Anthropic's next frontier model finally has a timeline. Chief Scientist Jared Kaplan revealed that Claude 4 is slated to arrive "in the next six months or so."

Outpacing the silicon cycle

Kaplan noted that AI development cycles are actively compressing, moving even faster than the underlying hardware cycle, despite constant arrivals of new chips.

According to Kaplan, the real momentum right now is coming from post-training and reinforcement learning. From where Anthropic sits, progress shows no signs of slowing down.

Why it matters

With Claude 3.5 Sonnet already serving as the daily driver for many software engineers and enterprise workflows, a rough six-month window gives teams clarity. You can keep building on the current stack without immediate fear of obsolescence, while budgeting for a major upgrade later this year.

Kaplan's comments also push back against the narrative that AI scaling is hitting a wall. By highlighting post-training and RL over raw compute, Anthropic is signaling that algorithmic gains are doing the heavy lifting, keeping the competitive heat turned all the way up.