AI model pricing is now a race to zero, and the labs charging premium prices will lose
Community vote
Probably True — 67% of 3 community votes. Official sources never set this.
Tried-it reports
Lived experience sits next to the community vote. Official sources never override either.
Tap a verdict. It counts right away, no sign-up needed.
Alibaba's Qwen4-Max launched overnight and devs on X keep pointing at the price: about $1.50 per million input tokens, the same entry price people are quoting for Google's new lightweight Gemini 4 model. The take going around is that frontier-ish AI is turning into a commodity and anyone charging 5-10x more is about to get squeezed. The counter is that cheap tokens don't matter if the model needs three retries, and that companies pay for reliability, support, and not having to explain a vendor choice to legal. Is AI pricing really a race to zero, or will the top labs keep their premium?
Votes and comments are community opinion, not professional advice.
More claims to call
b/ai →- A solo creator's AI 'Nuremberg' trailer just proved one person can out-hype a studio sci-fi trailerb/ai · Unrated · 1 comment
- ChatGPT fight picks are worthless, the Gurule upset at UFC 334 just proved itb/ai · Sketchy (1 vote) · 1 comment
- Gemini 4 Carbon means Google has finally passed OpenAIb/ai · Sketchy (1 vote) · 1 comment
- The Microsoft 365 outage proves Teams and Copilot still aren't ready for actual workb/ai · Unrated · 5 comments
Comments (1)
price per token is falling, total spend per company is going up. both can be true