If you've watched AI pricing lately, you've seen something remarkable: near-frontier capability now costs a fraction of what comparable quality did a year ago. Top models offered at half their tier's former price, frontier-grade models given away as free defaults, strong models priced at a few dollars per million tokens. This isn't a temporary promotion — it's structural. Here's why.
The forces driving prices down
Several compound at once:
- Competition — many labs now offer comparable top-end quality, so no one can charge a huge premium for long. Parity is weeks away, so price becomes the battlefield.
- Efficiency gains — better architectures, distillation, quantization, and inference optimization keep cutting the cost to serve a given capability.
- Open-weight pressure — strong, free open models (many from China) set a floor that closed providers must price against.
- Strategy — labs increasingly bet on volume, distribution, and ecosystem over premium margins. Cheap, widely-used models win mindshare and data.
A capability lead evaporates in weeks. A cost and distribution advantage compounds. So the smart money races the price down, not just the benchmark up.
What it changes for builders
Cheaper capability rewrites what's feasible. Features that were too expensive per call become viable. You can route more traffic to strong models. Long-horizon agents (which burn lots of inference) become economical. The old discipline of rationing your best model deserves a fresh look — the constraint loosened.
The caveat
Lower prices don't build products. Reliability, evaluation, cost management, good UX, and matching the right model to each task still decide success. Cheap intelligence lowers the floor; the work moves up the stack.
Why it matters
The collapse in capability-per-dollar is arguably the story of 2026 — bigger than any single model. It turns frontier intelligence from a scarce luxury into an abundant commodity, and pushes value toward products, data, reliability, and distribution. For anyone building, it's an invitation to be more ambitious about what's now affordable — because a lot more just became affordable.
