Google · Text models

Gemini 4 Hydrogen launches: Google's fast model for coding and agents

Google released Gemini 4 Hydrogen on Nov 12: GA in the Gemini API at $1.50 / $7.50 per million tokens, a new xhigh thinking level, and in the app and Search.

Google released Gemini 4 Hydrogen on November 12, 2026. Google describes it as a fast, efficient model “built for rapid execution across diverse agentic workflows”, aimed at everyday coding and agent work. It sits below the frontier models Gemini 4 Carbon and Argon, and Google says it can also act as the fast “execution engine” alongside them.

What to know

  • Status: the Gemini API release notes list gemini-4-hydrogen as generally available (GA), on both the free tier and paid pricing. Google says it is “rolling out globally”.
  • Price: Google lists $1.50 per million input tokens and $7.50 per million output tokens, flat across the full 1 million token context window, with output of up to 1 million tokens.
  • Thinking levels: four settings, low, medium, high (the default) and a new xhigh. See our thinking level prompts.
  • Where you can use it: the Gemini API and Google AI Studio (all tiers), Android Studio, Gemini for business, Google Antigravity for paid Google AI Pro and Ultra subscribers, the Gemini app model picker, and Google Search for all users (what changes in Search). Free Gemini app users get it through automatic model selection when it fits the prompt. Gemini in Google Sheets starts with AI Pro and Ultra subscribers and eligible business plans.
  • Antigravity: the release notes say Hydrogen is now the default model for the Antigravity agent in Gemini Managed Agents. Google says a new Orchestrator feature pairs Hydrogen with Carbon.
  • Speed (Google’s figures): Google says that at its default setting Hydrogen finishes tasks about 2.5x faster than Gemini 3.8 Flash, and that partners saw 2x to 2.8x faster runs with 12% to 23% fewer tokens.
  • Benchmarks (Google’s own figures): 51 on the Artificial Analysis Intelligence Index at the default level (Argon: 50), rising to 54 at xhigh; 57% on Terminal-Bench 4.1 against 18.2% for Gemini 3.8 Flash; 71.6% on Agents’ Last Exam v1.1.
  • Safety (Google’s figures): Google says Hydrogen makes more than 5x fewer fabricated claims than 3.8 Flash and is up to 2x less likely to bypass permission boundaries on blocked tasks. Google links a model card for details.
  • Migration: Google points developers moving from Gemini 3.8 Flash to its “What’s new” guide. Separately, Google has said gemini-2.5-pro and gemini-2.5-flash requests will be rerouted from December 1, 2026 (details).

Why it matters

Hydrogen is cheaper than Argon and Carbon ($2 / $10 and up) and Google claims it scores about the same as Argon on its headline index. For most everyday coding, drafting and agent steps, it is now the first Gemini 4 model to try. Our updated Gemini 4 model guide compares all three, and the Gemini page and model tracker have the wider picture.

Source: Google. We summarise in our own words; check the source for the full details.

Related

More AI news

stay in the loop

New guides, prompts and AI model news by email