Google Quietly Killed the Gemini Pro Free Tier. Here's What You're Actually Getting Now.

We logged into AI Studio last month and got the surprise: Gemini 3.1 Pro’s free tier—gone. No warning banner, no migration grace period. Just a “set up billing” prompt where the free credits used to be. According to Google’s official pricing docs, the paid tier is now the only way to access Pro-class models. If you’ve been prototyping with Gemini’s generous free tier, your side project just graduated to paid.

What Actually Died

Google’s Gemini API free tier used to look like this: unlimited tinkering with Gemini 2.5 Pro, rate-limited but real. Now it looks like this:

• Gemini 3.1 Pro / 2.5 Pro: Paid tier only. No free trial. No free playground tier anymore.
• Gemini 2.5 Flash / Flash-Lite: Still free, but the rate limits have a complicated history. In late 2025, Google cut Flash’s free tier hard—from around 250 requests per day down to 10–20 RPM on the strictest accounts—then walked it back to a formalized limit that’s tighter than the original but better than the crackdown floor.

The current official limit: Gemini Flash free tier maxes out at 1,500 requests per day and 15 requests per minute. That’s not useless—it’s fine for a prototype or a low-traffic integration—but it’s still a step down from the pre-2025 baseline.

The March and April Curveballs

Here’s the part that stung: as of March 23, 2026, Google switched to a prepay billing model. You buy credits up front ($10 minimum, $5,000 max). Those credits expire after 12 months. Then, on April 1, 2026, automatic monthly spend caps kicked in—Tier 1 accounts hit a $250/month wall, Tier 2 hit $2,000, and higher tiers climb from there. We missed the announcement on both. The spend cap isn’t theoretical; it pauses your API calls if you hit the limit mid-month.

For context on how we got here: the early generous limits were reportedly never intended to be permanent—one Google PM acknowledged they were supposed to be available for a single weekend, not two years. So Google tightened them.

The Math Right Now

Flash free tier:

• 1,500 requests per day
• 15 requests per minute
• Free forever (or until Google changes this again)

Pro paid tier (Tier 1 entry):

• No rate limits published upfront (they vary by tier and your request size)
• $0.075 per million input tokens, $0.30 per million output tokens (pricing varies slightly by model)
• $250/month spend cap, then API calls pause

If you’re calling Gemini Pro at scale, you’re hitting that cap. If you’re calling Flash at scale, you’ll either hit the rate limits or need to shard across multiple API keys. Neither is great.

Should You Stay on Gemini’s Free Tier?

Honestly: depends on the workload. We’re still using Gemini Flash free for throwaway prototypes and internal tooling where latency doesn’t kill us. The rate limits are tight, but 1,500 requests per day is real work if you’re smart about batching.

But if you need Pro-class reasoning or your project actually makes money, you’re paying now. The free tier was a Trojan horse—good enough to get you building, not good enough to ship on. We’ve moved some work to Claude Pro and OpenAI’s paid API where the cost-per-task makes sense, and we’re keeping a Gemini free key for the stuff that doesn’t need to be smart.

The prepay model is also a shift. You’re not paying for what you use month-to-month; you’re pre-funding a pool and hoping you don’t burn through it. Google’s not the only company tightening the screws on free tier—we’ve been watching this across the AI tool space. Expect more of this as the market matures.

The fix: if you’re building something real, get billing set up now. Flash free is still a fine prototyping tier, but don’t count on it staying generous. And if you were coasting on Pro free, your invoices just got a lot longer.