Google introduced Gemini 4 Hydrogen on November 12, 2026 for coding and AI-agent workflows. Alongside it, the company introduced Orchestrator in Antigravity, which switches between models such as Carbon and Hydrogen within the same task. The aim is to match model capabilities to individual stages rather than use one model throughout a workflow.
Dividing work between models
Google reports that on Terminal-Bench 4.0, Orchestrator achieved near-Carbon task success while consuming 24% less quota and finishing 19% faster on average. These are benchmark-specific results, not a guarantee of equivalent savings on every project. Lower quota consumption does not necessarily mean an identical reduction in an API bill.
Carbon can coordinate, while Hydrogen powers worker and verifier agents in isolated workspaces. This does not remove the need to define permissions and review changes.
Performance and pricing
Hydrogen scores 51 on the Artificial Analysis Intelligence Index at its default High thinking level and 54 at Xhigh, compared with 50 for Argon. A higher composite score does not establish superiority on every task.
The context window is one million tokens. Google specifies flat pricing across that window: $1.50 per million input tokens and $7.50 per million output tokens. Total task cost also depends on token volume, thinking level and retries.
Availability and practical evaluation
Google announced a global rollout through Gemini, AI Studio, Gemini API, business offerings and Search. Access terms differ by product; announced Antigravity access is for paid Google AI Pro and Ultra subscribers. Broad availability should not be confused with unlimited free usage across all surfaces.
Teams in Israel can test routing on a defined workflow, measuring time, quality, manual corrections and total consumption against a single-model run. The benefit depends on sensible handoffs and context preservation, not just Hydrogen’s benchmark score. All performance figures here are attributed to Google’s reporting.

