OpenAI has a problem, and for the first time in the company's history, the problem has a name: Anthropic.

Reports this week confirmed what insiders have been whispering for months. OpenAI internally declared "code red" against its biggest rival, launching a company-wide strategy to claw back the developer market that Anthropic has been steadily eating. The first visible result landed on April 9: a new $100-per-month ChatGPT Pro tier that offers 5x more Codex usage than the $20 Plus plan.

The price? $100. The same as Anthropic's Max 5x tier. The positioning? Targeted at developers. The same market Anthropic dominates. The strategy? If you cannot beat them, match them. Then outspend them.

Codex Hit 3 Million Users. That Is Not Enough.

Sam Altman announced that Codex, OpenAI's coding agent, reached three million users this week. On paper, that is impressive. In context, it is defensive. Claude Code has been eating Codex's lunch among professional developers for months. Anthropic's revenue just passed OpenAI's. And the developer community has been vocally switching to Anthropic's tools for anything involving serious code.

The new pricing structure reveals the panic. OpenAI now offers: a free tier, an $8 Go plan, a $20 Plus plan, a $100 Pro plan, and a $200 Ultra plan. That is five pricing tiers for a chat product. Anthropic has three. The complexity alone tells you OpenAI is trying to plug every hole in the dam at once.

The Super App Gamble

The pricing tier is just one piece. OpenAI also confirmed plans to build a desktop "super app" integrating ChatGPT, Codex, and its Atlas AI browser into a unified platform. The idea: instead of competing with Anthropic on individual tools, build an ecosystem so comprehensive that developers never need to leave.

It is the Microsoft playbook. Bundle everything. Make switching painful. Compete on convenience, not capability. The question is whether developers will buy it. The ones who switched to Claude Code did not switch because of price. They switched because Claude is better at writing code. A super app that contains a worse coding agent is still a worse coding agent.

The Spud Factor

There is one wild card that could change everything. Greg Brockman, who returned to OpenAI earlier this year, reportedly described an internal model codenamed "Spud" as a significant step toward artificial general intelligence. If Spud delivers, it resets the board. A model that represents a genuine capability leap would make the pricing war irrelevant.

But that is a big "if." OpenAI has promised capability leaps before. GPT-5 was supposed to be the moment OpenAI pulled away. It was good. It was not a generational leap. Meanwhile, Anthropic shipped Mythos and terrified the Treasury Secretary. The gap between what OpenAI promises and what Anthropic delivers is growing, and developers have noticed.

The Real Question Nobody Is Asking

Here is the uncomfortable truth about "code red." When the company that defined the AI industry has to declare an emergency against a competitor that did not exist four years ago, something fundamental has shifted. Anthropic is no longer the scrappy safety-focused alternative. It is the company that OpenAI is afraid of.

Matching Anthropic's pricing, copying Anthropic's developer focus, and building a super app to lock developers in are all reasonable moves. But they are defensive moves. And in a market moving this fast, playing defense is how you lose.

OpenAI does not need to match Anthropic's pricing. It needs to build something Anthropic cannot. So far, "code red" looks a lot like "code copy."