OpenAI has launched a new $100 per month ChatGPT subscription tier designed to compete with Anthropic’s Claude Code offering. The change centers on how much Codex usage subscribers can access, along with continued access to OpenAI’s “exclusive Pro model” and unlimited access to Instant and Thinking models—features OpenAI says are still part of the new Pro tier.
According to OpenAI’s announcement on X, the new Pro plan provides “5x more Codex usage than Plus” and is positioned as best for “longer, high-effort Codex sessions.” OpenAI is also running a time-limited promotion that increases Codex usage for eligible users until May 31, while it adjusts how Codex usage is allocated for Plus subscribers going forward. (See mint – technology for the full details, including the stated pricing and the promotion window.)
What OpenAI changed in ChatGPT Pro
The headline change is a new subscription price point: $100/month. OpenAI says this new Pro tier still includes access to all Pro features, including the exclusive Pro model. OpenAI also states that the tier provides unlimited access to Instant and Thinking models.
Where the tier differentiates is Codex usage. OpenAI says the new plan offers “5x more Codex usage than Plus.” In the same announcement, OpenAI frames the tier as suitable for “longer, high-effort Codex sessions.” That language suggests the company is shaping the experience around sustained coding workflows rather than short bursts, using usage limits as the mechanism to steer how people allocate time and compute for coding tasks.
OpenAI is also offering a launch promotion. In its post, the company says it is “increasing Codex usage for a limited time through May 31st”. The promotion is targeted at Pro subscribers: “Pro $100 subscribers get up to 10x usage of ChatGPT Plus on Codex” to help users “build your most ambitious ideas,” as OpenAI put it.
The promotion is time-bounded, and OpenAI says the Codex promotion for existing Plus members “will end today.” In addition, OpenAI says it is rebalancing Codex usage for Plus users to “support more sessions throughout the week, rather than longer sessions in a single day.” OpenAI’s stated framing indicates the company is not only changing total allowance tiers but also the distribution pattern of usage within a week.
Pricing and the rest of OpenAI’s ChatGPT tiers
OpenAI is not replacing its other plans. The company says it will continue to offer $200/month Pro alongside the $20/month Plus plan. It also continues to list an $8 “Go” plan and a free tier.
OpenAI explicitly characterizes the Plus plan at $20 as the “best offer” for “steady, day-to-day usage of Codex,” while describing the $100 Pro tier as a “more accessible upgrade path for heavier daily use.” These statements matter because they show OpenAI is drawing a ladder between tiers based on expected user behavior—daily usage patterns for Plus versus heavier daily use for the new Pro tier, with longer sessions supported by increased Codex allowance.
OpenAI CEO Sam Altman is also referenced in the same source. Altman had earlier announced that Codex had reached three million users, and that the company would reset usage for its users every million users. The mint – technology report links this context to the new subscription changes, placing them within an ongoing effort to manage Codex demand and usage accounting as the user base grows.
Why usage limits are becoming the battleground
This announcement reflects how AI coding tools are increasingly packaged as usage-based experiences. Instead of only differentiating models by capability, OpenAI is differentiating by how much Codex usage a subscriber can consume and how that usage is structured over time.
OpenAI’s own language shows two levers:
1) Total allowance by tier: The new Pro plan offers “5x more Codex usage than Plus.”
2) Temporal allocation: OpenAI says it is rebalancing Plus Codex usage to support more sessions throughout the week rather than longer sessions in a single day.
From a technology and product operations standpoint, these levers can affect compute scheduling, session planning, and how users design their coding workflow. The promotion—up to 10x usage for Pro $100 subscribers through May 31st—also indicates OpenAI can temporarily expand capacity or relax limits for a subset of users, then tighten back to the standard tier after the window closes.
OpenAI’s approach also ties the subscription directly to Codex usage rather than only to access to models. While OpenAI highlights unlimited access to Instant and Thinking models in the Pro tier, the primary “upgrade” metric presented in the report is Codex usage. That suggests Codex is the product component most sensitive to demand and thus most likely to be metered through subscriptions.
Competition with Anthropic: tier design echoes Claude Code
The mint – technology report notes that OpenAI’s subscription structure now looks similar to Anthropic’s. Specifically, it states that OpenAI’s plan “look[s] eerily similar to Anthropic,” describing Anthropic’s tiers as Max 5x for its $100/month users and Max 20x for its $200/month tier users.
OpenAI’s new tier provides 5x more Codex usage than Plus at $100/month, and the report frames this as part of OpenAI’s effort to rival Anthropic’s Claude Code popularity. The comparison matters because it shows how competitive pressure may push companies toward similar product packaging strategies—particularly when a key differentiator is the amount of coding-tool compute or usage a subscriber receives.
The report also links OpenAI’s subscription revamp to broader competitive context, including references to OpenAI executing a “code red” to counter Anthropic’s dominance in the coding market, and a shift toward more professional tool work. It further notes that OpenAI has put other plans on hold or shut them down, such as the recent Sora video generator (as described in the source material). While those points extend beyond subscriptions, they provide context for why OpenAI is focusing on coding-related tooling and on tier mechanics that map to developer usage.
As an industry signal, observers may watch whether usage-based tiering becomes a standard pattern for AI coding assistants—where the main product differentiation is how much “coding work” the subscription allows, and how that allowance is timed and reset as demand grows.
Source: mint – technology