Google's Gemini 4 Hydrogen speeds up coding, research, agentic workflows
What's the story
Google has launched its latest artificial intelligence (AI) model, Gemini 4 Hydrogen. The new system is said to be 2.5 times faster than its predecessor, Gemini 3.8 Flash. The move is part of Google's effort to enhance the speed and efficiency of everyday AI tasks such as coding and agentic workflows.
Model accessibility
Availability and focus on speed and efficiency
Gemini 4 Hydrogen is available on a number of platforms, including Google AI Studio, the Gemini API, Gemini for business, and the Gemini app.
Paid subscribers of Google AI Pro and Ultra can also access it via Google Antigravity.
According to Sundar Pichai, CEO of Google, this model was specifically designed for speed and efficiency in everyday tasks as well as coding and agentic workflows.
Cost structure
Thinking levels and pricing
Gemini 4 Hydrogen comes with four thinking levels, low, medium, high, and extra-high.
This lets users select the depth of reasoning according to their needs.
The model is priced at $1.50 per million input tokens and $7.50 per million output tokens across its one-million-token context window.
Benchmark success
Significant improvements over predecessor
In Google's evaluations, Gemini 4 Hydrogen more than tripled its score on the Terminal-Bench 4.1 coding benchmark compared to Gemini 3.8 Flash.
It scored an impressive 57% against Flash's 18.2%.
The new model also reduced fabricated claims by over five times, showing a significant improvement in performance and reliability over its predecessor.
Innovative feature
New orchestrator feature in Antigravity
Google is also launching a new feature called Orchestrator in Antigravity.
This tool automatically switches between AI models during a session based on the task at hand.
On Terminal-Bench 4.0, Orchestrator achieved task success rates similar to those of Gemini 4 Carbon while consuming an average of 24% less quota and completing tasks 19% faster.
Enhanced safety
Enhanced safety features in Gemini 4 Hydrogen
Gemini 4 Hydrogen can work with other models, with one handling coordination while others perform tasks such as coding, testing, and verification.
It is also built for research, data analysis, and document-related tasks.
Google has also improved safety in this model by reducing excessive actions in agentic tasks by up to 81% and making it less likely to bypass permission boundaries when faced with blocked or impossible tasks.
Twitter Post
Official post
Meet Gemini 4 Hydrogen.
— Google (@Google) November 12, 2026
It brings cutting-edge intelligence to everyone, with speed and efficiency for everyday coding and agentic work.
Across 4 thinking levels, it stands on its own or teams up with Carbon and Argon — 2.5X faster than 3.8 Flash, outpacing competing models in… pic.twitter.com/QTRwACdxU2