Gemini 4 Hydrogen Is Google's New Fast Executor, Priced Flat at $1.50 In and $7.50 Out Across a 1M-Token Window

Google has released Gemini 4 Hydrogen, a faster and cheaper member of the Gemini 4 family built for everyday coding and agentic workflows. The model was announced on November 12, 2026 by Koray Kavukcuoglu, SVP at Google DeepMind and Google’s Chief AI Architect. Where Gemini 4 Carbon and Gemini 4 Argon are positioned for deep, long-horizon reasoning, Hydrogen is pitched as the model that does the execution work: writing code, running tests, checking changes and following instructions across many tool calls.

Hydrogen is also the release that puts Gemini 4 in front of everyone. Google is rolling it out across all tiers of Google AI Studio and the Gemini API, Gemini for business, the Gemini app, Google Search, and Google Antigravity for paid Google AI Pro and Ultra subscribers.

One price across the full million tokens

Hydrogen is priced at $1.50 per million input tokens and $7.50 per million output tokens, and Google says that rate is flat across the entire 1 million-token context window rather than stepping up for long prompts. The model accepts up to 1 million input tokens and can produce up to 1 million output tokens.

It comes with four thinking levels: low, medium, high (the default) and a new xhigh tier. Google describes the levels as a smooth trade-off between quality, latency and cost, so developers only pay for the reasoning depth a task needs.

On the Artificial Analysis Intelligence Index, Google says Hydrogen scores 51 at its default high setting, one point above Gemini 4 Argon’s 50, and reaches 54 at xhigh.

Built to pair with Carbon and Argon

The release is framed around multi-model work. Google expects teams to keep using Carbon and Argon for architectural decisions and subtle bugs, while Hydrogen handles the faster loop of edits and verification.

To automate that split, Google introduced Orchestrator in Google Antigravity. Instead of one model running a session from start to finish, Orchestrator switches between models such as Carbon and Hydrogen inside the same session. On Terminal-Bench 4.0, Google says Orchestrator reached near-Carbon task success while using 24% less quota on average and finishing 19% faster. For parallel agent setups, Carbon can act as coordinator while Hydrogen powers worker and verifier agents in isolated workspaces.

Numbers against Gemini 3.8 Flash

For teams on Gemini 3.8 Flash, Google is selling Hydrogen as a direct upgrade. At its default setting the company says Hydrogen finishes tasks 2.5 times faster, more than triples Terminal-Bench 4.1 results (57% versus 18.2%) and cuts fabricated claims by more than five times. Early partners working on graph, SQL and data-heavy enterprise workloads reported Hydrogen running 2x to 2.8x faster with 12% to 23% fewer tokens than 3.8 Flash.

Google also lists results on agentic and domain benchmarks: 58.8% on HLE-Diamond, where it says Hydrogen leads its class; 52.3% on Vals Tax Agent Bench, ranked first overall; 55.8% on Vals Legal Research Bench, tied for first; 53.4% on the new Multimodal Knowledge Agent benchmark; and 71.6% on Agents’ Last Exam v1.1. It says Hydrogen ranks fifth on arena.ai’s live Agent Arena. All of these figures are Google-reported.

Safety claims for an agent model

Because Hydrogen will run many unattended tool calls, Google spent a section of the launch on agent behaviour. Compared with 3.8 Flash, it says Hydrogen reduces overreach in agentic tasks by up to 81% and under-service by up to 79%, and cuts misaligned actions and task shortcutting across tool-use and coding evaluations by 12x. When a task is blocked or impossible, Google says Hydrogen is up to two times less likely to cross permission boundaries and scores 96.7% on an internal agent honesty evaluation.

Google also reports 99.9% chain-of-thought legibility with no deceptive reasoning edits across more than 200,000 audited turns, and says Hydrogen is more than 2.3x more resilient on the Gray Swan indirect prompt injection benchmark. False refusals on HealthBench Pro are kept under 0.4%, according to the company.

Where it shows up for consumers

In the Gemini app, Hydrogen appears in the model drop-down, and free users get it through automatic model selection when Google’s router decides it fits the prompt.

Search is getting Hydrogen for all users globally. Google says it handles multi-part and typo-heavy queries better, returns more visual answers and upgrades the query fan-out technique to surface a wider range of links. In Google Sheets, Gemini with Hydrogen is available first to Google AI Pro and Ultra subscribers and to eligible business plans, with Google claiming outputs more than 30% faster than previous models in the Sheets canvas.

For developers, Hydrogen is available now in the Gemini API through AI Studio and Android Studio, and in Antigravity for paid subscribers. Enterprises can select it in Gemini for business through the API or the model picker.