Anthropic and OpenAI spent months warning the world about the existential risks of advancing AI too quickly. They signed letters. They testified before Congress. They looked very concerned in Bloomberg interviews.
Now they've released cheaper models.
OpenAI dropped GPT-6 Sol and GPT-6 Luna. Anthropic countered with Claude Opus 5.5. Both companies slashed prices to undercut each other. The apocalypse can wait. Market share cannot.
This marks their first major release since the slowdown letter made headlines. That letter argued AI development posed catastrophic risks requiring immediate regulation and voluntary restraint. Hundreds of researchers signed it. The media covered it extensively. Politicians nodded thoughtfully.
Then someone at OpenAI checked the revenue projections.
The cheaper models target enterprise customers tired of paying premium rates for chatbots that hallucinate legal citations. Both companies framed the releases as democratizing access to advanced AI. They did not mention that democratizing access directly contradicts calling for slower development to prevent societal collapse.
Anthropic's press release never used the word slowdown. OpenAI's blog post focused on efficiency gains and cost savings. Neither company explained how racing to release discounted models aligns with their previous stance that this technology requires unprecedented caution.
Retail traders immediately began speculating on which company would dominate the cheaper model segment. They created spreadsheets comparing per-token pricing. They argued about benchmarks in Discord servers. Not one of them stopped to consider that both companies might be full of sh*t about the existential risk thing.
The slowdown lasted exactly as long as it took for someone to blink first on pricing.
Photo by on Unsplash
Leave a Comment