Gemini 4 Hydrogen: pricing, benchmarks and how it compares

Gemini 4 Hydrogen at a glance

What's new in Gemini 4 Hydrogen

• A fast model in the Gemini 4 family: Google keeps Carbon and Argon for deep, long-horizon reasoning and positions Hydrogen for rapid execution, on its own or as the worker beside them.
• Orchestrator in Google Antigravity switches between Carbon and Hydrogen in one session. Google says it reached near-Carbon success on Terminal-Bench 4.0 while using 24% less quota and finishing 19% faster.
• A big step up from Gemini 3.8 Flash, per Google: 2.5 times faster at the default high setting, Terminal-Bench 4.1 up from 18.2% to 57%, and more than five times fewer fabricated claims.
• Batched exploration of large repositories: it reads and searches files in parallel rather than one by one. Partners saw it run 2 to 2.8 times faster with 12% to 23% fewer tokens than 3.8 Flash.
• Four thinking levels, including a new xhigh, so you pay for only the reasoning a task needs.
• Safety in agentic work: Google reports up to 81% less overreach, 12 times fewer misaligned actions and shortcuts in tool-use and coding evaluations, and a 96.7% honesty rate on its internal agent evaluation.

Gemini 4 Hydrogen benchmarks

These are the numbers Google published in its announcement; independent results will follow. Where Google gave a comparison, it's in the last column.

Gemini 4 Hydrogen pricing

Gemini API prices per million tokens in USD, from Google's pricing page on November 13, 2026. Output prices include thinking tokens. Carbon and Argon have launch prices until November 30, then double; 3.8 Flash's price doubles on January 1, 2027. Claude prices are Anthropic's, from our Claude Fable 5.2 post.

Gemini 4 Hydrogen vs Argon vs Carbon

Google splits the work by depth. Carbon is its most capable reasoning model, for complex software engineering, deep math and science, and autonomous multi-step agents. Argon is the efficient frontier model, with faster responses and lower token spend than Carbon. Hydrogen is for fast, everyday coding and agentic work, and it can run as the worker or verifier while Carbon coordinates.

On price, Hydrogen costs less than Argon and Carbon today and the gap grows on December 1, when their launch prices end. On Google's numbers it also edges past Argon on the Artificial Analysis Intelligence Index (51 vs 50 at high). Neither Carbon nor Argon has a free tier in the Gemini API; Hydrogen does.

Gemini 4 Hydrogen vs Gemini 3.8 Flash

If you run on 3.8 Flash, Google pitches Hydrogen as the upgrade: faster at its default setting, more than three times the Terminal-Bench 4.1 score and far fewer fabricated claims. It costs twice as much per token as 3.8 Flash until December 31; from January 1, 2027 the two list at the same $1.50 and $7.50. Google's partners also report 12% to 23% fewer tokens per task, which offsets part of the difference.

Gemini 4 Hydrogen vs Claude Sonnet 5.5 and Opus 5.5

Google's announcement doesn't compare Hydrogen with Claude, so here are the facts that can be compared. At $1.50 and $7.50 per million tokens, Hydrogen lists below Claude Sonnet 5.5 ($2 and $10) and well below Opus 5.5 ($4 and $20), and it offers a 1 million-token context with up to 1 million output tokens. Claude Code and Codex run their own vendors' models, so the choice of model usually follows the agent you work in; for how those two compare, see Claude Code vs Codex.

How to use Gemini 4 Hydrogen

1. Try it in Google AI Studio

   Pick Gemini 4 Hydrogen in AI Studio; it's available on every AI Studio tier, including free.

2. Call it from the Gemini API

   Use the model code gemini-4-hydrogen. Leave thinking at the default high, and move to medium or low when speed matters more than depth.

   # Gemini API: model code gemini-4-hydrogen
   curl "https://generativelanguage.googleapis.com/v1beta/models/gemini-4-hydrogen:generateContent" \
     -H "x-goog-api-key: $GEMINI_API_KEY" \
     -H "Content-Type: application/json" \
     -d '{"contents": [{"parts": [{"text": "Find and fix the failing test in this diff"}]}]}'

3. Code with it in Google Antigravity

   Paid Google AI Pro and Ultra subscribers get it in Antigravity, where Orchestrator can switch between Carbon and Hydrogen in one session.

4. Use it in other agents

   Coding agents that accept a Gemini API key can call gemini-4-hydrogen once they list the model; check your agent's model list.

Source: Google, Introducing Gemini 4 Hydrogen

A faster model is half of it. The other half is a job.

Models keep getting faster and cheaper, and the agents around them still wait for the next prompt. Crewmark turns your coding agent sessions into AI employees with a role, a goal in numbers, working hours and a heartbeat, so they take the next issue from a board on their own.

1. Pick a role

   From the AI employee roles, such as an AI software engineer.

2. Put the work on a board

   Assign issues, make the regular ones repeat, and let the heartbeat wake the employee to take the next one.

3. See what it can do

   Jobs with setup and rules for each agent are in Claude Code use cases and Codex use cases.

FAQ