OpenAI launches $100 ChatGPT Pro tier with 5x more Codex usage

• OpenAI added a new $100/month ChatGPT Pro tier that, according to OpenAI's announcement, includes all Pro features and 5x more Codex usage than Plus.
• The old $200 Pro plan stays in place as the highest-capacity option, and OpenAI's follow-up says existing $200 subscribers keep a temporary 2x Codex promo through May 31.
• Plus is getting a limit rebalance rather than a straight increase, with OpenAI's thread reply saying the plan now favors more sessions spread across the week instead of longer bursts in a single day.
• OpenAI's own help docs now frame Codex usage as token-based credit consumption, and the Codex rate card says fast mode burns 2x credits.
• The weirdest tell is capacity pressure: after another rate-limit reset, Thomas Sottyaux said current growth would force yet another reset in under two weeks while OpenAI scrambles to bring more capacity online.

You can see the new slot on OpenAI's pricing page, read the terse community announcement, and dig into the newly updated Pro plan help page. The more revealing docs are the new Codex rate card, which switches to token-based metering, and OpenAI's credits FAQ, which says Plus and Pro users can now buy extra Codex usage instead of upgrading plans.

OpenAI finally filled the gap between Plus at $20 and Pro at $200. The company says the new $100 tier is meant for longer, higher-effort Codex sessions, while still bundling the same broad Pro feature set, including the Pro model plus unlimited access to Instant and Thinking models.

The pricing page now shows Pro as a distinct step above Plus, while the updated About ChatGPT Pro plans page says all Pro plans include Codex, deep research, image creation, memory, and file uploads. In practice, the big differentiator is usage allowance, not feature access.

OpenAI is also treating launch week like a promotion. Thomas Sottyaux's clarification says both the $100 and $200 Pro plans are getting an extended 2x limits promotion, and his main launch post says OpenAI reset limits again at rollout.

The company did not position this as a Plus upgrade. According to OpenAI's thread reply, the existing Codex promotion for Plus subscribers ended on April 9, and the plan is being rebalanced to support more sessions throughout the week rather than long single-day runs.

That matters because it changes the shape of the quota more than the sticker price. A Plus user who occasionally camps in Codex for a half day may feel tighter ceilings, while a user making smaller daily runs may see less interruption. The Hacker News discussion picked up on exactly that distinction, with commenters arguing the launch is less about new features than about packaging more reliable coding throughput for heavier users.

The docs around this launch are more interesting than the tweets. OpenAI's Codex rate card now describes Plus, Pro, Business, and Enterprise/Edu usage with token-based credits instead of rough per-message averages, explicitly tying cost to input, cached input, and output tokens.

A few details stand out:

• Fast mode consumes 2x credits.
• Code review uses GPT-5.3-Codex.
• OpenAI says average Codex cost lands around $100 to $200 per developer per month, with wide variance by model, automations, and fast mode usage.
• The separate credits FAQ says Plus and Pro users can buy pay-as-you-go credits after hitting included plan limits.

That makes the new $100 tier look less like a clean new bundle and more like one rung in a broader flexible-pricing system for Codex-heavy accounts.

The most concrete operational signal came a couple of hours after launch. Sottyaux said Codex demand is growing fast enough that OpenAI may owe users another rate-limit reset in less than two weeks, and joked about resets happening more than once a week before the product reaches 10 million while the team rushes capacity online.

That lines up with the rest of the rollout. Sam Altman framed the $100 plan as a response to popular demand, but the surrounding evidence points to a second constraint: OpenAI is simultaneously segmenting heavier coding users by willingness to pay and buying time while infrastructure catches up.