More Room to Work With Gemini Models
If you've been bumping up against usage caps in Google AI Studio, you'll want to know about the changes that landed on April 20. Google expanded the rate limits across several Gemini models for subscribers on its AI Pro and AI Ultra plans — and the difference between tiers is pretty significant.
AI Pro runs $19.99/month and now gives you up to 50 prompts per day for Gemini 3.1 Pro, plus 100 prompts for Nano Banana Pro. On top of that, Pro subscribers get a $10 monthly credit toward Gemini API usage. Not life-changing on its own, but it adds up if you're prototyping regularly.
AI Ultra — at $249.99/month — is a different animal entirely. Ultra users get substantially higher allowances, including 6,250 daily prompts for Nano Banana 2. That's the kind of headroom you need when you're running serious workflows, not just poking around.
Developer Tools Are Getting a Meaningful Upgrade
Here's where it gets interesting for developers specifically. It's not just about raw model access anymore — Google is pushing its suite of developer tools harder with these updated limits.
AI Pro subscribers now get 5x higher limits for Jules, Google's asynchronous coding agent. You also get higher daily model requests for both Gemini Code Assist and Gemini CLI. And Google Antigravity — the coding agent that came out of Google's $2.4 billion acquisition of the Windsurf team — gets "Higher" agent request allowances under Pro as well.
Ultra subscribers get even more. We're talking 20x higher Jules limits, specifically designed for what Google calls "intensive, multi-agent workflows at scale." The CLI and Code Assist tools hit the "Highest" daily request tier too. If you're orchestrating multiple agents simultaneously, this is the tier built with you in mind.
The Free Tier Is Quietly Shrinking
And here's the part that deserves a bit of attention. These expanded paid limits didn't happen in a vacuum — they're part of a broader shift in how Google is monetizing its AI platform.
On April 1, Google pulled Pro-series Gemini models from the free tier of the API entirely, making them paid-only. Flash-series models still run at no cost, but the more capable stuff is now behind a paywall. Google also introduced mandatory prepaid billing for new users back in March, along with enforced monthly spending caps tied to billing tiers.
That last part — the spending caps — came out of a pretty rough situation. In late 2025, a billing system incident left some developers staring at charges north of $70,000 for services they hadn't actually used. Google responded by rolling out project-level spend caps in AI Studio on March 16. So the guardrails are tighter now, which honestly feels overdue.
The core AI Studio interface itself remains free for prototyping and prompt design. But if you're moving toward production use through the Gemini API or Vertex AI, you're looking at a more structured cost environment than a year ago.
Vibe Coding Gets a Real Home
One more thing worth flagging: this update quietly cements AI Studio as a serious environment for what the industry is calling "vibe coding" — the workflow where you describe what you want in plain language and let the AI handle the actual implementation. Google launched this in AI Studio in March 2026 through the integrated Antigravity agent, and the higher limits under paid plans make it a lot more practical for sustained sessions.
It's a calculated move. Give developers the tools and the headroom, and they start building production workflows inside your ecosystem. That's the play here.