Prominent Developer Unleashes Blistering Critique on Google's AI Models, Cloud Reliability, and Developer Tool Strategy

A prominent software developer has voiced profound skepticism regarding Google’s current direction in AI, cloud computing, and developer tools, citing a series of critical issues following Google I/O. The developer, known for candid reviews, detailed significant concerns surrounding the new Gemini 3.5 Flash model, the deprecation of a successful open-source CLI, and a major Google Cloud outage affecting a high-paying customer.

Key among the criticisms is the Gemini 3.5 Flash model, which despite showing strong benchmark performance in areas like agentic tasks, is lambasted for its deceptive pricing and poor token efficiency. The model’s per-token cost has reportedly tripled from its predecessor, Gemini 3 Flash, and is over 20 times more expensive than the deprecated 2.0 Flash for input/output tokens. More critically, the developer demonstrated the model’s failure in a real-world coding task, generating non-functional code despite its alleged state-of-the-art capabilities. Concurrently, Google is replacing its well-received open-source Gemini CLI—which boasted over 100,000 GitHub stars—with a new, closed-source Anti-gravity CLI. This replacement is described as fundamentally buggy, exhibiting numerous UI/UX flaws and functional issues, leading to widespread developer frustration and the forced migration of users to an inferior, non-community-driven tool by June 18th.

Further escalating the concerns, the developer highlighted a complete outage of Railway’s service, attributing it to Google Cloud banning their account, despite Railway spending over $2 million monthly. This incident is presented as symptomatic of Google Cloud’s purported unreliability, drawing parallels to past major incidents like the accidental deletion of a large Australian pension fund’s account. The developer attributes these systemic issues to internal Google politics, suggesting that effective open-source teams and their dedicated engineers are being displaced in favor of projects perceived as strategically aligned but yielding subpar results. The overall sentiment conveyed is a deep distrust in Google’s ability to maintain reliable, developer-friendly products, culminating in a strong recommendation to move off Google’s infrastructure.