Google’s Gemini API gets two new tiers: Flex for cheap, Priority for speed

Google just dropped two new inference tiers for the Gemini API: Flex and Priority. The pitch is straightforward — you pick how much speed you need and pay accordingly.

Flex is the budget option. It routes your requests through lower-priority queues, so you get cheaper inference but with no latency guarantees. Think of it as batch processing for real-time-ish apps that can tolerate a few extra seconds. If you’re building a summarization tool where results don’t need to appear instantly, Flex makes sense.

Priority is the opposite. It reserves capacity upfront, ensuring faster response times. You pay a premium, but you get consistent latency — useful for chat interfaces, live agents, or anything where users are waiting on a response.

This isn’t revolutionary pricing — cloud providers have been doing similar tiering for years — but it’s a welcome addition for anyone who’s been burning money on Gemini’s standard tier when they didn’t need to. I’ve personally seen projects where the cost of Gemini was the main reason teams switched to cheaper models. Flex might keep them on the platform.

What I’d like to see next: clearer documentation on how much slower Flex actually is under real-world load. Google says “no guarantees,” but developers need ballpark numbers to decide. Also, Priority’s pricing needs to be competitive with the standard tier, or it’s just a rebranding exercise.

That said, this is a step in the right direction. Not every use case needs sub-second responses, and not every budget can afford them. Giving developers a toggle between cost and speed is better than a one-size-fits-all API.