I've been using Cursor for months now to build UserJot with Claude Sonnet and Opus, and it's been pretty solid. But I'm starting to hit its limits more often.

Claude Code keeps popping up everywhere. My Twitter feed is full of developers talking about it. Friends are switching over. Even my most stubborn vim-using colleagues are saying the $200 Max plan is worth it for the higher usage limits, way more than what you get with Cursor's Ultra tier at the same price.

And you know what? Cursor seems worried. They just rolled out new pricing tiers: Pro ($20), Pro+ ($60), and Ultra ($200) with varying usage limits, clearly trying to keep up. But Claude Code's Max plan offers significantly more usage, and more importantly, access to Opus 4 which is just better at coding than anything else I've tried.

I was literally planning to give Claude Code a try this week. Then Google dropped a bomb.

Here's the pricing breakdown that changes everything:

• Gemini CLI: Free (60 requests/min, 1,000/day)
• Claude Code Pro: $20/mo (10-40 prompts/5 hours)
• Claude Code Max: $100/mo (100-200 prompts/5 hours)
• Claude Code Max: $200/mo (200-800 prompts/5 hours)
• Cursor Pro: $20/mo (~225 Sonnet 4 requests/mo)
• Cursor Pro+: $60/mo (~675 Sonnet 4 requests/mo)
• Cursor Ultra: $200/mo (~4,500 Sonnet 4 requests/mo)

Enter Gemini CLI: Google's "Hold My Beer" Moment

Recently, Google announced Gemini CLI and the pricing is absolutely wild:

• 60 requests per minute
• 1,000 requests per day
• Completely free (Just sign in with a personal Google account to get started)

Google says they actually measured typical developer usage, then doubled it to set these limits. They're betting most devs won't even come close.

For context, Claude's Pro plan ($20/mo) gives you roughly 10-40 prompts every 5 hours. The Max plan at $200/mo gets you 200-800 prompts in the same timeframe.

Google is basically saying: "Why pay $200/month when you can get more for free?"

Of course, there's a catch for enterprises. If your company needs multiple agents running in parallel, or has specific data residency requirements, you'll need to spring for a paid license. But for individual developers? It's genuinely free.

It's Not Just About Price

Here's the thing: this isn't some half-baked tool Google threw together. Gemini CLI runs on Gemini 2.5 Pro, which has been quietly becoming developers' favorite model. The 1 million token context window means you can throw entire codebases at it without breaking a sweat. Plus, it's multimodal, so you can feed it PDFs, sketches, or images. And while the model runs in the cloud, the CLI itself is local.

They've also made it properly open source (Apache 2.0), integrated it with Google Search for real-time data, and built it to handle more than just coding, from automating PR reviews to general research and task management. It even hooks into Google's other AI tools like Imagen and Veo for media generation. And if you're a VS Code user, it integrates seamlessly with Gemini Code Assist. Basically everything Claude Code has, minus the price tag.

What This Really Means

Google is clearly playing catch-up here. While Anthropic has been steadily building a loyal developer following with Claude Code, Google's been watching from the sidelines. Now they're using their classic playbook: enter the market late, but with such aggressive pricing that it forces everyone to reconsider.

The timing is interesting too. Just as Claude Code was gaining serious momentum with their integrated Pro/Max plans, Google drops this free alternative. It's the same strategy they used with Gmail vs Hotmail, Chrome vs Firefox, and countless other products.

My Take

Am I switching from Cursor? Yes, but the question is to what. Gemini CLI's free tier is hard to ignore, but I'm actually leaning towards Claude Code's Max plan. Here's why: Opus 4 is just better at coding. It understands context better, writes cleaner code, and catches edge cases that other models miss.

I'm building UserJot, a feedback, changelog, and product roadmap tool that helps teams stay connected with their users. Right now I'm deep in features that help companies collect feedback, manage their roadmaps, and publish beautiful changelogs. And honestly? Having reliable AI coding assistance makes a huge difference when you're shipping features constantly.

What's fascinating is watching AI pricing collapse in real time. These models have been heavily subsidized by VCs, often operating at a loss. Now with open source LLMs getting better and developers building tool alternatives that work with any model, the competition is fierce.

Just look at the last few months: Cursor's new three-tier structure (matching Claude's $200 top tier), Claude's Max plan, and now Google's free offering. It's a race to the bottom in AI pricing, and developers are winning.

For those already paying for Claude Code, Google just made that monthly expense look pretty questionable. Though for me, the quality difference with Opus 4 might still justify the cost.

The good news? All this competition means better tools and pricing for developers. Whether you go with Claude's superior models, Gemini's free tier, or stick with Cursor, we have more options than ever.

Now if you'll excuse me, I need to try out both Gemini CLI and Claude Code.