Changelog

Gemini 3.8 Flash retires onto Gemini 4 Hydrogen

Gemini 3.8 Flash is no longer offered, and everything that still named it now runs on Gemini 4 Hydrogen, which already took over our own defaults from it. A specialist, workflow, conversation, or default model set to 3.8 Flash moves to Hydrogen with nothing to change on your side. So do older picks of Gemini 3.5, 3.6 and 3.7 Flash and Gemini 3.1 Pro, which had been running as 3.8 Flash. Compare Hydrogen with the other hosted models on the models page.

How it works

  • Price. Hydrogen costs $1.50 input and $7.50 output per million tokens, twice 3.8 Flash's $0.75 and $3.75, and cached input stays at $0.075 per million. Hydrogen writes fewer tokens per task, and Google raises 3.8 Flash itself to Hydrogen's price on January 1.
  • Effort carries over. A saved medium or high runs at that level on Hydrogen, which also offers low. A level Hydrogen doesn't offer, such as minimal from an older Flash, runs at its default, high.
  • Workspace settings carry over. A workspace that turned 3.8 Flash off has Hydrogen off too, unless it had already set Hydrogen itself. A workspace whose utility model was 3.8 Flash now uses Gemini 3.5 Flash Lite, our default for background work.
  • Subscriptions. A Gemini plan connected through Token Tunnel no longer lists 3.8 Flash, and a pick saved as 3.8 Flash runs as Hydrogen when the plan offers it.

Also shipped

  • A hosted chat completion request that names 3.8 Flash is answered and billed as Hydrogen.

Your first specialist starts here.

Give it one job. Build the workflows, apps, and skills to get it done.