Google launches Gemini 4 Hydrogen, its faster lightweight AI model
Hydrogen completes tasks 2.5 times faster than Gemini 3.8 Flash and is available now through Google AI Studio, the API, the Gemini app and Search.
Google released Gemini 4 Hydrogen, a model optimized for fast processing and AI agent tasks, according to aitimes.com. The company announced the model on its official blog on November 12 and began offering it through Google AI Studio, the API, the Gemini app and Search.
Hydrogen is the successor to Gemini 3.8 Flash and was developed based on Gemini 4 Carbon. Google says it completes tasks 2.5 times faster than the previous lightweight model. It supports up to 1 million tokens of context and accepts text, images, audio and video as input to generate text, with a default maximum output of 64,000 tokens that can be expanded to 1 million in a separate long-generation mode.
Benchmark results
Hydrogen scored 51 on the Artificial Analysis Intelligence Index at the High reasoning setting, above Argon's 50, and 54 at the Xhigh setting. It recorded a 57% success rate on Terminal-Bench 4.1, against 18.2% for Gemini 3.8 Flash, and 76.1% on AutomationBench-AA, above Claude Sonnet 5.5's 71.8% and GPT-6.1 Terra's 60.2%. It also scored 71.6% on the Agents' Last Exam 1.1 benchmark.
The model offers four reasoning levels: Low, Medium, High and Xhigh. API pricing is $1.50 per million input tokens and $7.50 per million output tokens.
Google also introduced an Orchestrator in its Antigravity AI coding tool that switches automatically between Carbon and Hydrogen depending on the task. With Orchestrator enabled on Terminal-Bench 4.0, quota usage fell by an average of 24% and completion time by 19% while success rates stayed close to Carbon-only use.
Box CEO Aaron Levie said agent accuracy rose from 74% with Flash to 79% with Hydrogen, with task time cut by more than half. In Zapier's independent evaluation, Hydrogen recorded a 51.1% task success rate, nearly matching Argon's 51.3%, at about half the cost per task.
Hydrogen is the third new Gemini 4 model, following Carbon, which is specialized for complex reasoning, and Argon. Gemini 3.8 Flash currently costs $0.75 per million input tokens and $3.75 per million output tokens, and will move to the same pricing as Hydrogen from next January; Argon and Carbon had introductory launch pricing of $2 per million input tokens and $10 per million output tokens.
Quick answers
Where can I use Gemini 4 Hydrogen?
Google is offering it through Google AI Studio, the API, the Gemini app and Search.
How much does Gemini 4 Hydrogen cost through the API?
API pricing is $1.50 per million input tokens and $7.50 per million output tokens.
What is the context limit for Gemini 4 Hydrogen?
It supports up to 1 million tokens of context, with a default maximum output of 64,000 tokens that can expand to 1 million in a separate long-generation mode.