AI agents · OpenClaw · self-hosting · automation

Quick Answer

What Is Gemini 4 Hydrogen? Price, Access and Benchmarks

Published:

The short answer

Gemini 4 Hydrogen is the fast, cheaper member of Google’s Gemini 4 family, released November 12, 2026. It costs $1.50 per million input tokens and $7.50 per million output with no long-context surcharge, has a 1 million-token context window with up to 1 million output tokens, and is the first Gemini 4 model on the free API tier. Model ID: gemini-4-hydrogen.

Price and access

Tier (per 1M tokens)InputOutputCached input
Standard (paid)$1.50$7.50$0.075
Batch / Flex$0.75$3.75$0.0375
Priority$2.70$13.50—
Free tierFreeFreeFree (data used to improve products)

Implicit caching only; explicit caching is not supported. Grounding with Google Search: 5,000 free requests a month shared across Gemini 3 and newer models, then $14 per 1,000. Source: Google’s Gemini API pricing page, read November 15, 2026. All model prices are on our current API prices table.

Where you can use it: Google AI Studio and the Gemini API (all tiers), Android Studio, Gemini for business (API and model picker), Gemini in Google Sheets for eligible plans, the Gemini app model picker, Google Search for all users, and Google Antigravity for paid Google AI Pro and Ultra subscribers.

Where it sits in Google’s lineup

ModelRolePrice nowAfter the intro period
Gemini 4 CarbonFrontier, deep reasoning$2 / $10$4 / $20 from Dec 1, 2026
Gemini 4 ArgonFrontier$2 / $10$4 / $20 from Dec 1, 2026
Gemini 4 HydrogenFast everyday and agentic execution$1.50 / $7.50Flat (no intro period announced)
Gemini 3.8 FlashPrevious speed tier$0.75 / $3.75 (intro)$1.50 / $7.50 from Jan 1, 2027

From December 1, 2026, Hydrogen will cost less than half of Carbon or Argon per token, which is why Google pitches it as the “execution engine” next to a frontier planner.

What Google claims

All figures are Google’s, from the November 12, 2026 launch post:

  • Artificial Analysis Intelligence Index: 51 at default high thinking (Argon 50), 54 at xhigh.
  • Terminal-Bench 4.1: 57% vs 18.2% for Gemini 3.8 Flash; tasks finished 2.5× faster than 3.8 Flash.
  • Partners: 2–2.8× faster and 12–23% fewer tokens than 3.8 Flash on graph, SQL and data-heavy workflows.
  • Other results: HLE-Diamond 58.8%, #1 on Vals Tax Agent Bench (52.3%), #5 on Arena’s Agent Arena, Agents’ Last Exam v1.1 71.6%.
  • Safety: up to 81% less overreach in agentic tasks and 2.3× more resilient to indirect prompt injection on the Gray Swan benchmark.

Four thinking levels: low, medium, high (default) and a new xhigh. Pay only for the depth a task needs — thinking tokens are billed as output.

Orchestrator: Hydrogen plus Carbon

Google launched Orchestrator in Google Antigravity alongside Hydrogen. It switches between Carbon and Hydrogen within one session; on Terminal-Bench 4.0 Google reports near-Carbon success rates with 24% less quota and 19% faster completion. For parallel agents, Carbon coordinates and Hydrogen powers worker and verifier agents. The same pattern works outside Antigravity: plan with a frontier model, execute with a cheaper one.

Should you switch?

  • From Gemini 3.8 Flash: yes for agentic and coding work; test cost per task, because Hydrogen is 2× the per-token price until 3.8 Flash’s own increase on January 1, 2027.
  • From Argon for routine work: likely, before Argon’s December 1 doubling.
  • Versus other mid-tier models: compare against Claude Sonnet 5.5 and GPT-6.1 Terra on your own tasks — see Gemini 4 Argon vs Claude Sonnet 5.5 vs GPT-6.1 Sol for the tier above, and what is Gemini 4 Carbon.

Last verified: November 15, 2026.

Sources