OpenAI and Anthropic Cut Prices on the Same Day. Your AI Budget Just Changed.

Last Monday, something unusual happened. OpenAI and Anthropic, the two biggest names in AI, slashed their API prices within 90 minutes of each other. That's not a coincidence. That's a price war. And if your business spends money on AI tools, you just got a raise.

What Actually Happened

OpenAI dropped GPT-6 Sol and Luna on September 22. Sol runs at $2 per million input tokens and $10 per million output. Luna costs a dime per million input. Both are 50% cheaper than their predecessors in the GPT-5.6 line.

About an hour and a half later, Anthropic fired back. They launched Claude Opus 5.5 with pricing 20% below Opus 5 and claimed it costs 40% less to run.

Neither company released a more powerful model. They released cheaper ones. That tells you where the fight is right now.

Price Drops Change the Math

New AI models get all the headlines. Price cuts change the spreadsheet.

When OpenAI sold GPT-5.6 at $4 per million input tokens, plenty of businesses looked at the bill and decided to limit how they used it. They'd run the expensive model on high-value tasks and use something cheaper for everything else.

Sol cuts that cost in half. Luna drops it to basically nothing. A business running 10 million API calls a month just saw their bill go from "we need to talk about this" to "keep going."

And Anthropic's move means you're not locked in. Two top-tier providers are actively trying to undercut each other. That pressure only goes one direction for buyers.

The Three-Tier Play

OpenAI now has three models at three price points:

Astra ($10/$50 per million tokens): The big brain. Complex reasoning, research, strategy work. You use this when you need the model to think hard.

Sol ($2/$10): The workhorse. Coding, agent tasks, business workflow automation. This is where most companies will spend most of their budget.

Luna ($0.10/$0.50): The assembly line. Document extraction, ticket summarization, classification. Tasks you need to run thousands of times a day without watching the meter.

This gives businesses a clear decision tree. Match the model to the job. Stop paying Astra prices for Luna work.

The Timing Tells You Everything

Ten days before these price cuts, Anthropic CEO Dario Amodei publicly called for slowing down AI capability gains. Sam Altman and Elon Musk both agreed within hours.

Then both companies shipped cheaper, more efficient models. Not more powerful. Just cheaper.

Look at those two facts side by side. The industry's most prominent voices said "let's slow down." Then they made it cheaper for everyone to speed up.

I don't think that's hypocrisy. I think it shows you where the real competition is. The frontier model race has cooled off. Cost is where the fight is now.

What to Do About It

If you're already using AI, audit your costs this week. You might be running a $4 model where a $2 model does the same job. That's a 50% savings you can capture by changing one line of code.

If you've been holding off because AI API costs felt unpredictable, the floor just dropped. Luna-tier pricing means you can automate high-volume, low-complexity tasks for almost nothing.

If you run an AI agency or consultancy, your margins just improved. The tools cost less. Your expertise is worth the same.

The people who benefit most from a price war are the ones actually building with these tools.

— Mark Garza, Laimen AI