Home Process Plans About
From the blog

The AI Model War Is a Good Thing for Your Business (Here's Why)

September 23, 2026 · 4 min read

OpenAI launched GPT-6 Sol and Luna this week with a permanent 50% price cut. Anthropic responded within 48 hours with Opus 5.5. A coordinated industry slowdown on new releases collapsed.

If you run a small or mid-size business, this looks like noise. More models. More announcements. More pressure to figure out which one to bet on.

Subscribe now

Here’s the reframe: this is the best thing that could have happened for you. Not because the models are getting better - they are, but that’s not the point. Because the competition is now about price, not capability. And when the race shifts from “who can build the smartest model” to “who can deliver the most value at the lowest cost,” the buyer wins every time.


The 4-Month Head Start Rule

The most important data point of this entire launch cycle came from Ars Technica, not from any press release. Their reporting found that paying 5x the cost for access to frontier AI models only buys about a 4-month head start over cheaper alternatives.

Four months. That’s the premium you’re paying for. Not permanent advantage. Not exclusive capability. A four-month window before the same capability arrives at a fraction of the price.

For a startup racing to build a product around a specific model, that head start might matter. For a service business trying to automate customer support or document processing, it almost certainly doesn’t.

The gap between what expensive and cheap models can do is shrinking fast. And with GPT-6 Sol and Luna now pricing below Opus 5.5 and DeepSeek’s V4.1 Flash, the market is effectively telling you: you don’t need to chase the frontier. The frontier is coming to you.

The Coordination Ended

One thing the press coverage nailed this week: the AI industry’s coordinated push for a slower, safer, more deliberate release cadence just fell apart. OpenAI and Anthropic released competing models in the same week, directly contradicting their own public calls for caution.

This matters for business buyers because it means the pace of improvement isn’t slowing down. Every quarter, there will be a new model that does more for less. That creates a natural temptation to wait - to hold off on implementation because something better is always just around the corner.

That temptation is the real trap.

If you’re waiting for the model landscape to stabilize, you’ll be waiting forever. The pace isn’t going to slow. What will happen is that the cost of being wrong about which model you chose will continue to drop. Pick a good enough model today, build your workflow around it, and if a better one arrives in four months - great. The switching cost is lower than you think, because the models are converging on the same basic capabilities.

What This Means for Implementation

Here’s the practical takeaway: the hardest part of AI adoption was never picking the right model. It was always doing the work to fit the tool into your actual business processes.

The model war makes that even more true. When the underlying technology is getting cheaper and more capable every quarter, the differentiation isn’t in which model you use. It’s in how well you integrate it into the work your team actually does. The implementation is the moat. Not the model.

This is where the 15% vs. 40% productivity spread that McKinsey tracked comes into play. Companies that treat AI as a license to activate - turn it on, send an email, hope for the best - end up at the low end of that spread, regardless of which model powers their tool. Companies that map the tool to real workflows, test it with one person first, and maintain their automations over time end up at the high end.


The Strategy Shift

The AI model war is telling you something useful: stop optimizing for the wrong variable.

Don’t spend your time comparing benchmark scores between GPT-6 Luna and Opus 5.5. Spend it figuring out which three tasks in your business consume the most manual effort and would benefit most from automation. The model that handles those tasks well enough exists today. And if a better one ships next quarter, you’ll swap it in without missing a beat, because the workflow - not the model - is what’s delivering the value.

This is the moment to be a fast follower, not a frontier chaser. Let the model companies fight over the bleeding edge. You can capture most of the value at a fraction of the cost by implementing well on last-generation technology.

At Black&Tan Labs, we see this play out every engagement. The businesses that get the most out of AI aren’t the ones that pick the newest model. They’re the ones that do the work to fit it into how their team operates. That’s a Discovery engagement - a full day in-person, at $10k, mapping exactly where AI pays off in your business and where it doesn’t. For teams already running automations, the Care & Feeding retainer at $500/month per automation keeps everything current as the technology underneath keeps shifting.

The model war is good for your business. Don’t watch it. Use it.

Enjoyed this? New essays on AI, operations, and building in contrast - delivered to your inbox.

Subscribe on Substack →