Google Introduces Gemini 4 Hydrogen, Built for Fast Everyday Coding and Agentic Workflows

Google introduced Gemini 4 Hydrogen in November 2026, presenting it as a model with speed and efficiency for everyday coding and agentic workflows. Google detailed the model in a blog post and in posts from its Google AI and Google Gemini accounts on X. Google separates the model from its frontier models, Gemini 4 Carbon and Gemini 4 Argon, which it says excel at deep, long-horizon reasoning on hard problems. Hydrogen, by contrast, is built for rapid execution across diverse agentic workflows, the company says.

What Google Said in Its Posts on X

In a post on X, the Google AI account said Hydrogen beats Argon at its default thinking level on the Artificial Analysis Intelligence Index, where Google's blog gives scores of 51 against 50. The same post put its speed at 2.5 times that of Gemini 3.8 Flash. The Google Gemini account said on X that the model is better at calling the right tools and gets users to a reliable output with less back-and-forth. It also said Hydrogen follows instructions more closely.

Thinking Levels, Pricing and Context Window

Hydrogen has four thinking levels: low, medium, high as the default, and a new xhigh tier. Google says the Intelligence Index score reaches 54 when the model runs at its highest xhigh thinking setting. Input tokens cost $1.50 per million and output tokens cost $7.50 per million, a flat rate across the 1 million-token context window. The model card says Hydrogen accepts text, images, audio and video, with a combined input-output context window of up to 1M tokens. According to the card, Hydrogen is adapted from Gemini 4 Carbon.

Orchestrator and Multi-Model Coding Workflows

Google also introduced Orchestrator in Google Antigravity, which switches between models such as Carbon and Hydrogen in one session. On Terminal-Bench 4.0, Google says Orchestrator reached near-Carbon task success while using 24% less quota and finishing 19% faster on average. For parallel-agent workflows, Google says Carbon can coordinate while Hydrogen runs worker and verifier agents that build and test changes in isolated workspaces. Google's developer post lists interactive web development, multimodal workflows and repository coding and debugging among its strengths.

Benchmarks and Gains Over Gemini 3.8 Flash

Google presents Hydrogen as a significant standalone upgrade for teams moving from Gemini 3.8 Flash. It reports a Terminal-Bench 4.1 score of 57% for Hydrogen against 18.2% for that model. Partners testing complex graphs, SQL and data-heavy enterprise workflows saw Hydrogen run 2x to 2.8x faster with 12% to 23% lower token usage than 3.8 Flash, Google said. On deep domain reasoning, Google says Hydrogen leads its class on HLE-Diamond with 58.8% and ranks first overall on Vals Tax Agent Bench at 52.3%. It ties for first place on the Vals Legal Research Bench with 55.8%. Google reports 53.4% on its new Multimodal Knowledge Agent benchmark and 71.6% on Agents' Last Exam v1.1. On Agent Arena, arena.ai's live benchmark, Hydrogen ranks fifth, according to Google. Google's evaluation notes say results for non-Gemini models come from providers' self-reported numbers unless otherwise stated. For Claude Sonnet 5.5 and GPT-6.1 Terra, the document says Google defaults to the maximum thinking or reasoning settings available. Hydrogen's own results are run through the Gemini API at the highest thinking settings unless indicated otherwise, the document adds.

Availability, Search and Safety Notes

Google says Hydrogen is rolling out globally across its developer, enterprise and consumer products. Developers can use it through the Gemini API in Google AI Studio and in the paid tier of Android Studio. Paid Google AI Pro and Ultra subscribers can try it in Google Antigravity. Enterprises can deploy it in Gemini for business through the API or the model drop down, and in Gemini in Google Sheets for eligible business subscriptions. In Google Search, Hydrogen is available globally for all users. Google calls the change the biggest model upgrade to Search yet. Google says Hydrogen improves the query fan-out technique in Search, linking users to a broader range of sites. Google says Hydrogen cuts overreach in agentic tasks by as much as 81% and under-service by as much as 79%. The model card lists hallucinations among Hydrogen's known limitations. It gives the model's knowledge cutoff as March 2026. The model card says a full Safety and Security Report for Hydrogen will follow in the coming days.