Google's AI pricing triples annually. That single fact reframes the entire competitive landscape. While Google's Gemini 2.1 Pro still undercuts rivals at $2.00 input and $12.00 output per million tokens, the trajectory is aggressive. OpenAI's GPT-5.5 sits at $5.00 input and $30.00 output. Anthropic's Claude Opus 4.7 lands at $5.00 input and $25.00 output. Google is cheaper, but closing that gap fast.
Tomasz Tunguz traces three distinct strategies through the pricing history charts. OpenAI subsidized its flagship model for a period, then raised prices as margins tightened. Anthropic held luxury pricing steady before cutting its most powerful models late last year. Google kept raising across the board while staying low in absolute terms. The patterns reveal when each company prioritized growth over profit and when that calculus flipped.
All three vendors are now raising or holding prices as capex spending hits record levels. The subsidy era is ending. The original piece is worth reading for the historical pricing charts alone, which show exactly when each company blinked. The next question is whether Google's tripling annual rate signals a floor approaching, or a ceiling still being built.
[READ ORIGINAL →]