Gemini vs. Claude: A Google Developer Expert’s Open Letter to Google Cloud
I love Gemini 3 at home, but I’m being forced to build my professional future on Claude Code.
To my fellow devs and the Google Cloud Product teams,
For the past few months, my development workflow has been split down the middle.
In my spare time, I am all-in on the Google ecosystem. I’ve been using gemini-cli, pushing Gemini 3 to its limits on advanced personal projects, and the experience has been great. The CLI is maturing at a rapid pace, and the AI Pro plan is one of the best value-for-money moves I’ve seen in the space (and as a GDE, I’m grateful for the access).
But there is a growing friction point that is becoming impossible to ignore. When I clock in for my day job, that experience disappears and it’s pushing me (and many other enterprise devs) toward the competition.
Why global only is failing the EU
Here is the irony: at my job, we are a GCP shop. But because of EU compliance and data residency requirements, we are stuck. Currently, Gemini 3 is only offered via the Global endpoint in Vertex AI. At work, we are restricted to regional endpoints. This means while I’m using Gemini 3 at home, I’m stuck with Gemini 2.5 at the office. And let’s be honest, 2.5 isn't anywhere close to the capabilities of the models we now see in the wild.
The model garden irony
This is where the frustration peaks. As a Google Cloud customer, I see the Vertex AI model garden acting with incredible speed for third-party models.
When Anthropic releases a model like Claude 4.6, Google Cloud has it deployed with EU regional endpoints within the same week. Yet, Google’s own flagship, Gemini 3, has been available since late November and still hasn’t made it to the regional endpoints in the EU.
It is a strange state of affairs when it is easier for an EU enterprise to use the latest Claude model on Google Cloud than it is to use the latest Gemini model.
The gravitational pull of Claude Code
Because of this gap, I’ve been using Claude Code at work. I don’t buy into all the hype, I still think Gemini 3 is the context king, but the tool is impressive. More importantly, it’s available where I work and I’m noticing a drift. Because I’m forced to use Claude Code for eight hours a day at the office, those habits are bleeding into my personal projects. Even with my GDE status and my preference for Gemini’s context window, I find myself gravitating toward the Claude ecosystem simply because the professional friction of Gemini is too high.
The path forward
I’ve heard the promising rumors about Gemini 3.5, and I want to be excited. But rumors don’t solve compliance hurdles. To keep developers like me from drifting “permanently”, we need two things:
Regional Parity: GCP must prioritise getting Gemini 3 (and all future flagship versions) into regional EU endpoints alongside the Global release. The “Regional Lag” is actively losing you enterprise adoption.
Early Access for Advocates: As a GDE, I want to help bridge this gap. Give us preview access to Gemini 3.5 so we can stress-test the next generation of agentic workflows before we’ve already settled into a competitor’s CLI.
I want to keep building with Gemini. I want the gemini-cli to be my primary tool. But in the current model war, availability is just as important as capability.
Let’s close the gap before the drift becomes a permanent shift.