Discover the Best AI Tools
We test the top AI tools for writing, video, images & music — so you don't have to.
Explore AI ToolsOpenAI Just Launched a $500/Month ChatGPT Plan — and It's All About Speed


OpenAI's DevDay 2026 keynote (Fort Mason, San Francisco, Sept 29) was about AI agents — but the pricing story may matter just as much. OpenAI unveiled Ultrafast, a premium speed tier, and a brand-new $500-a-month subscription called ChatGPT Pro 500 to go with it. Speed, it turns out, is now OpenAI's most expensive feature.
What Ultrafast actually is
Ultrafast is not a new model — it's a speed tier. OpenAI advertises up to 8x faster token generation in Codex, hitting around 300 tokens per second, and up to 6x faster through the API. On stage, Altman demoed it by building and launching a small rocket scene side-by-side: the Ultrafast version finished visibly earlier. GPT-6 Astra Ultrafast is available now; GPT-6.1 Sol Ultrafast is "coming soon." API pricing is premium too: Astra Ultrafast runs $60 per million input tokens and $300 per million output — six times the standard rates.
The $500 plan that comes with it
In ChatGPT, Ultrafast is locked behind a new top plan: ChatGPT Pro 500 at $500 a month. OpenAI says it carries the company's highest included usage allowance — 25 times ChatGPT Plus — plus access to Dots, the new always-on AI agents, and computer use through the Agents API inside ChatGPT Work. The Pro ladder is now Pro 100 ($100), Pro 200 ($200), Pro 500 ($500), and only Pro 500 includes Ultrafast.
The catch for the $200 tier
The same keynote quietly rewrote the $200 plan. OpenAI reopened Pro 200 to new subscribers — but new subscriptions now get half the old allowance: 10 times Plus instead of 20 times, and weekly GPT-6 Pro messages cut from 200 to 100. Existing subscribers keep their old limits until October 29, 2026. To soften the blow, OpenAI permanently removed the old five-hour usage window, so subscribers can use their weekly cap whenever they want instead of timing sessions around resets.
Why it matters
OpenAI is segmenting its users by speed. For developers and researchers where latency is money, 300 tokens a second is a genuine productivity upgrade — and Pro 500 is clearly built for them. But the downgrade of Pro 200 is a reminder that premium AI subscriptions now behave like cable bills: prices move in both directions, usually up. If you don't need bleeding-edge speed, the lower tiers are unchanged.
My take? Speed tiers are a clever business move — latency is the one bottleneck no prompt trick can fix. But watch the fine print: the most important DevDay number might be the $200 plan's shrinking allowance, not the $500 headline.
Comments
Post a Comment