Just days after overhauling its pricing structure, Amazon Web Services (AWS) is facing backlash from developers over a bug in its new agentic AI-driven integrated development environment (IDE), Kiro. The glitch caused usage limits to drain much faster than expected, sparking widespread frustration and criticism.
AWS Acknowledges the Bug
AWS confirmed that the issue stemmed from a bug introduced during Kiro’s revised pricing rollout. According to Adnan Ijaz, director of product management for Agentic AI at AWS, the problem caused some tasks to inaccurately consume multiple requests, leading to developers hitting their usage caps prematurely.
In a message posted on Kiro’s official Discord channel, Ijaz assured users that AWS was “actively” working on a fix and expected to resolve the issue within a couple of days. Meanwhile, the company is offering to reset limits for affected users.
A Troubled Pricing Rollout
Kiro’s pricing has been controversial since the IDE entered public preview. In July, AWS had already placed temporary limits on usage after overwhelming demand. Initially, AWS announced a straightforward three-tier pricing model:
Free: 50 agentic interactions per month
Pro ($19/month): Up to 1,000 interactions
Pro+ ($39/month): Up to 3,000 interactions
However, last week AWS shifted to a new pricing structure based on vibe and spec requests, unique to Kiro’s spec-driven development approach:
Free: 50 vibe, 0 spec requests
Pro ($20/month): 225 vibe, 125 spec requests
Pro+ ($40/month): 450 vibe, 250 spec requests
Power ($200/month): 2,250 vibe, 1,250 spec requests
Beyond these limits, AWS charges $0.04 per vibe request and $0.20 per spec request.
This new model has left many developers frustrated. Users have complained on social media, GitHub, and forums that their credits are disappearing too quickly, even in cases where they did not actively engage with the IDE.
Developer Backlash and Comparisons to Cursor
The reaction has been reminiscent of Cursor’s pricing controversy in June, when sudden changes confused users and led to unexpected cost spikes. Like Cursor, AWS now faces backlash from developers who are questioning Kiro’s value against its competitors.
Some developers have voiced concerns on GitHub, reporting that Kiro consumes a large number of vibe credits without active engagement. Others are worried about the long-term affordability of the tool given the unique and more complex pricing model.
Why Kiro Costs More Than Rival IDEs
Despite developer dissatisfaction, some analysts argue that Kiro may deliver more long-term value compared to competitors such as Cursor or Claude Code.
According to Jason Andersen, principal analyst at Moor Insights and Strategy, Kiro’s spec-driven approach differentiates it from conventional chat-based coding assistants. Instead of generating small code snippets, developers can define an entire application or task in one go.
However, this approach is also resource-intensive. A single spec request can spawn multiple tasks simultaneously, consuming significantly more GPU inference power — and therefore driving up costs.
“AWS introduced this split pricing model to balance affordability with the power of spec-driven development,” Andersen explains. “But it also means spec requests are far more expensive. To win developers over, AWS will need to educate them on what to expect in terms of usage and costs.”
The Road Ahead for AWS Kiro
AWS is working to restore confidence in Kiro by fixing the bug and resetting impacted developer limits. However, the bigger challenge may lie in winning back developer trust after multiple pricing shifts and confusing credit consumption rules.
For now, Kiro remains a promising but expensive bet for developers who want more than traditional AI code assistants can offer. If AWS can resolve technical bugs and better explain its vibe and spec pricing, Kiro could establish itself as a leading agentic IDE — but only if it can regain developer goodwill.
