You have 1 article left to read this month before you need to register a free LeadDev.com account.
Estimated reading time: 4 minutes
Just as AI coding assistants become deeply embedded in software engineering workflows the vendors are increasing prices and setting rate limits. What can be done?
Agentic IDEs and AI copilots are increasingly popular with professional software engineers and indie developers. However, tools like Claude Code, Replit, and Cursor are still maturing and poorly communicated changes to pricing policies have been met with significant blowback.
For dev teams that have implemented workflows built around these apps, these changes set an unwanted precedent. If prices and feature availability can change drastically overnight for key tools, how can you rely on them going forward?
Bait and switch
Take Claude Code. In July it cut usage limits for customers without any prior communication. A week later, it announced the new usage limits, including a weekly limit for the first time. Instead of having to wait a few hours for the cap to reset, heavy users (some of which have been optimizing their sleep patterns accordingly) could be cut off from Claude Code for a few days.
“Claude Code has experienced unprecedented demand since launch. We designed our plans to give developers generous access to Claude, and while most users operate within normal patterns, we’ve also seen policy violations like account sharing and reselling access, which affects performance for everyone,” an Anthropic spokesperson told LeadDev.
Your inbox, upgraded.
Receive weekly engineering insights to level up your leadership approach.
“To keep the service reliable, we’re adding weekly usage limits to Pro and Max plans starting August 28. Most users won’t notice a difference – the weekly limits are designed to support typical daily use across their projects. These changes will affect less than 5% of Pro and Max plan subscribers, and Max users can still purchase additional capacity at standard API rates when needed.”
Similarly, in June this year Cursor shifted from a request-based pricing model – where most users got 500 external model requests per month – to a usage-based pricing model, where every request was instead charged using API pricing with $20 of credit included with the pro plan. This was all despite some features being listed as “unlimited”.
Many users were unhappy with the changes and sudden increase in their bills. One user on Reddit reported going from around $100 per month to $20 to $30 every day without changing how they used Cursor.
To Cursor’s credit, they apologized and offered to refund affected users. But they also clarified that while a subset of features have unlimited usage, the most advanced AI models were now on usage-based pricing and that wasn’t going to change.
And it’s much the same situation with Replit. In June, it changed a vibe coding feature called Agent to “effort-based pricing”. Originally, users were charged $0.25 for any edits made by Replit’s Agent – with large tasks able to trigger multiple charges. Under the new effort-based pricing, some edits can cost less than $0.25 but other changes that require more AI usage now cost multiples of what they did previously, according to users on Reddit.
More like this
Business as usual
While Dr. Christina Inge, an instructor in AI in Marketing for Harvard University’s continuing education division, expresses sympathy for individual users, she feels it’s disingenuous for companies with professional managers to be too surprised by these kinds of price hikes. After all, many of these same companies pursue similar approaches to user acquisition and pricing.
“Freemium and penetration pricing are both pretty classic pricing models,” says Dr. Inge. The idea with these aggressive pricing models is to get people to try your product by offering it for an unsustainably low fee, and then when you get traction and users relying on it, increase the price to the point that you can be profitable. It’s the same pricing strategy that has been used by companies like Uber, DoorDash, Canva, and many others.
“The cost for these AI coding platforms to the consumer has always been artificially low,“ explains Dr. Inge. And that’s especially true of the power users most affected by changes like Claude Code’s usage caps, and Cursor and Replit’s usage-based pricing models.
One Claude Code user told TechCrunch that before the new limits, he had been able to get the equivalent of more than $1,000 worth of API usage in a single day with the $200 per month Max plan.
For engineering leaders considering whether to invest in an agentic IDE for their organization and, if so, which one, Dr. Inge’s advice is pretty clear cut: “No one gets fired for buying IBM” – or in this case major cloud providers like Microsoft, Google, and Amazon Web Services. As she explains it, the more stable the company, the more likely the pricing won’t increase aggressively without warning.
October 15-17, 2025
Big names joining the New York lineup! 🔥
For engineering leaders who have been hit with a price hike that they are unhappy with, Dr. Inge’s advice is to switch sooner rather than later. Vendor lock-in only gets worse the longer you stay with a platform so the cost of switching now is almost certainly lower than the cost of switching later.
And if you are switching, go with a big, popular, and stable option. That minimizes the risk of having to switch again further down the line. While the fast growing venture-backed agentic IDEs and vibe coding tools might seem exciting, they’re the ones most likely to struggle to find a profitable pricing model and pull the rug.
