OpenAI has unveiled two tiered versions of GPT-6, named Sol and Luna, priced to compete hard against Google's Gemini models and Anthropic's Claude. The move is a clear signal that the AI price war, which has been grinding costs down steadily since 2023, has entered a sharper, faster phase. What was enterprise pricing twelve months ago is now within reach of a ten-person business in Leith or a GP practice in Inverness.

The competitive pressure driving this is real and structural. Google has been cutting Gemini API prices repeatedly throughout 2024 and into 2025, while Anthropic has positioned Claude's mid-tier models as a cost-effective alternative for professional workflows. Alongside both of them, open-source models like Meta's Llama series have given developers a credible free route, forcing every commercial player to justify their price tag with better performance or lower cost. OpenAI is doing both at once.

According to reporting by AI Business, Sol is positioned as the more economical of the two new models, designed for high-volume, lower-complexity tasks, while Luna targets more demanding reasoning and generation work at a price point still well below previous GPT-4 class models. That kind of tiering matters for SMEs: you no longer have to pay for a sledgehammer when a tack hammer will do. You pick the model that fits the job.

The practical implication for Scottish businesses is straightforward. Research from the Federation of Small Businesses Scotland has consistently shown that cost and complexity are the two biggest barriers to AI adoption among SMEs. Price is now rapidly ceasing to be the obstacle. The complexity barrier is falling too, as tools built on these models, everything from AI receptionist software to automated bookkeeping assistants, become more capable and easier to configure without a developer on the team. Scottish Enterprise's AI Adoption Framework, published as part of the broader Digital Economy Strategy, identifies this exact window as critical: businesses that build working AI habits now will have a compounding advantage over those who wait until the tools are ubiquitous.

There is also a wider strategic point worth noting. As commercial AI providers race to the bottom on price, the value shifts from the raw model to whoever uses it best. A 30-person accountancy firm in Edinburgh that has already integrated AI into its client reporting workflow will not be disrupted by cheaper models becoming available; it will simply get faster and cheaper to run. The price war is good news for early adopters and a shrinking window for everyone else sitting on the fence.