“The excuse that AI is too expensive just ran out of shelf life.”

For years, the answer to “we can't afford AI” was a reasonable one. This week, it stopped being true.

An 80% price cut

OpenAI dropped the price of its GPT-5.6 Luna model by 80%, from around $1 per million input tokens down to $0.20. A million tokens is roughly 750,000 words, so that's the cost of some serious AI work landing at a fraction of what it cost a month ago. In the same announcement, OpenAI said its models now reach over one billion people and more than two million businesses.

The part that caught my attention

Here's the detail that actually caught my attention, not the price cut itself, but why it happened. Part of the saving came from OpenAI's own model, GPT-5.6 Sol, rewriting and optimising the GPU code that runs its infrastructure, cutting serving costs by 20% on its own. The tool got cheaper partly because the AI was put to work making itself cheaper to run.

The shelf life on that excuse

That's the part small businesses should sit with. The excuse “AI is too expensive for something my size” had a real shelf life. That shelf life is closing fast, and it's closing because the economics keep compounding in one direction.

To be fair, price was never the only barrier. Knowing which task is actually worth automating still takes real thought, and that thought is worth more than the token cost either way. But if cost was the reason you were waiting, it's worth checking the numbers again. They've probably changed since the last time you looked.

If you want help figuring out which task is actually worth automating for your business, that's exactly the kind of conversation worth having. Find me through the contact section of this site.