As a developer, I’m no stranger to API limitations. But when I came across OpenAI’s Tier 1 API limits, I was shocked. The token-per-minute cap is so low that it’s essentially impossible to run a single request using the Roo system prompt without hitting the ceiling. This isn’t just a minor inconvenience – it’s a complete blockade on meaningful development or testing.
I’m not alone in my frustration. Many developers are struggling to work with these restrictions, and it’s affecting their ability to build and innovate. It’s not just about being able to send a large number of requests; it’s about being able to send a single request that’s moderately sized.
The current limits are crippling, and they’re not just affecting hobbyists. Professional developers who are trying to build complex applications are being held back by these restrictions. It’s not just about being able to send a lot of requests; it’s about being able to send a single request that’s meaningful.
I understand that OpenAI wants to prevent abuse, but there’s a big difference between abuse prevention and hobbling legitimate users. Right now, Tier 1 is not just low – it’s non-functional for any serious use case.
OpenAI needs to rethink its Tier 1 restrictions and raise the token-per-minute limit to accommodate at least a single full request using a reasonable context size. They also need to make the tier progression to Tier 2 more transparent and attainable for verified, paying users. And most importantly, they need to stop conflating ‘security’ with ‘artificial starvation of resources.’
Until these changes are made, OpenAI is alienating the very developers who could be generating revenue and building the tools that showcase their platform’s strengths. Instead, they’re forcing us to either give up or move to competitors who aren’t actively sabotaging our ability to spend money on their services.
It’s time for OpenAI to listen to its users and make some changes.