The disappearing AI middle class
The price-performance curve just snapped—developers must pick a side now.
Last week, OpenAI and DeepSeek made opposite bets on the value of frontier AI. OpenAI doubled down on a closed, premium product with higher prices, while DeepSeek pushed open infrastructure with dramatically lower costs. The result is a price gap between the two ends of the market that is now wider than it has been in years, effectively splitting the model landscape into two clusters.
Until last week, developers could pick a model on a fairly smooth price-performance curve with a top tier, a middle tier, and a budget tier. That continuous gradient has now snapped. The comfortable middle that most coding agents routed through is thinning out. Developers building agents, coding assistants, and high-volume inference pipelines now face a binary choice: pay a premium for closed performance or go cheap with open infrastructure. This shift forces harder decisions about routing, cost, and capability trade-offs.
- OpenAI raised prices on its closed frontier models while DeepSeek slashed costs on open infrastructure in the same 24-hour window.
- The price gap between top-tier and budget-tier models is now the widest it has been in years, eliminating the middle ground.
- Developers of coding agents and high-volume inference pipelines must now choose between premium closed performance and cheap open infrastructure.
Why It Matters
Developers lose the middle-ground model option, forcing binary cost-performance trade-offs in agent and inference design.