
Gemini 4 Hydrogen: Flash-Tier Price, Argon-Level Smarts
Gemini 4 Hydrogen beats Argon on the Artificial Analysis index at $1.50 per million input tokens and runs 2.5x faster than 3.8 Flash. What changes for you.
Google's newest model is not its biggest one, and that is exactly the point. On November 12, Google DeepMind introduced Gemini 4 Hydrogen, a fast, low-cost model built for everyday coding and agentic work that, by Google's own numbers, edges past the larger Gemini 4 Argon on a popular composite intelligence benchmark. For most developers and Gemini app users, Hydrogen is now the Gemini 4 model they will touch most often.
- Price: $1.50 per million input tokens and $7.50 per million output tokens, flat across the full 1 million-token context window
- Headline score: 51 on the Artificial Analysis Intelligence Index at its default high thinking level (Argon scores 50), rising to 54 at the top xhigh setting, according to Google
- Speed: finishes tasks 2.5 times faster than Gemini 3.8 Flash and more than triples its Terminal-Bench 4.1 score (57% vs. 18.2%)
- Availability: live now in the Gemini API, Google AI Studio, Gemini for business, the Gemini app and AI Mode in Search
What Is Gemini 4 Hydrogen?
Hydrogen replaces the old "Flash" slot in Google's lineup. 9to5Google notes that Google dropped the Flash branding this time, and the model picker in the Gemini app now labels Hydrogen as the "Everyday help" choice, sitting below Argon for code and analysis and Carbon for the most complex work. Free Gemini app users reach it through the Auto model selection when it fits the prompt.
The design goal is disciplined execution. Google describes Carbon and Argon as the models for deep, long-horizon reasoning, while Hydrogen is tuned to write code, run tests and verify changes quickly. It takes up to 1 million tokens of input and can produce up to 1 million output tokens, which makes long multimodal reports over PDFs, spreadsheets, audio and video practical in a single pass.
How Does Hydrogen Compare With Gemini 3.8 Flash?
The upgrade over the previous fast model is large on paper. Beyond the speed and Terminal-Bench gains, Google says Hydrogen cuts fabricated claims by more than five times and recovers cleanly when a tool call fails or a search comes back empty. Early partners, Google reports, saw Hydrogen run 2x to 2.8x faster than 3.8 Flash while using 12% to 23% fewer tokens on SQL, graph and data-heavy enterprise workflows.
Google also lists several benchmark placements, all of which are the company's own claims rather than independent results:
- 58.8% on HLE-Diamond, leading its model class
- First overall on Vals Tax Agent Bench (52.3%) and tied for first on Vals Legal Research Bench (55.8%)
- 53.4% on the new Multimodal Knowledge Agent benchmark and 71.6% on Agents' Last Exam v1.1
- Fifth place on arena.ai's live Agent Arena leaderboard
On the safety side, Google says Hydrogen reduces agentic overreach by up to 81% versus 3.8 Flash, is up to 2x less likely to bypass permission boundaries on blocked tasks, and scored a 96.7% honesty rate on Google's internal agent-honesty evaluation.
Orchestrator: Carbon Plans, Hydrogen Does the Work
The most interesting launch detail for builders is Orchestrator in Google Antigravity. Instead of one model running a whole session, Orchestrator lets Gemini 4 Carbon hand routine steps to Hydrogen. On Terminal-Bench 4.0, Google says this setup reached near-Carbon success rates while using 24% less quota and finishing 19% faster. For parallel agent workflows, Carbon can coordinate while several Hydrogen workers build and verify changes in isolated workspaces.
This is the clearest sign yet that the frontier is becoming a team sport between models. A big reasoning model plans, a cheap fast model executes, and the bill shrinks. If you have been following our Gemini 4 Carbon API coverage, Hydrogen is the missing worker tier that makes that architecture affordable.
What It Means for Developers and Everyday Users
At $1.50 in and $7.50 out, Hydrogen undercuts the intro pricing we covered in our Gemini 4 Argon pricing breakdown while, per Google, matching Argon on the composite index. For teams running coding agents all day, that is a meaningful cost lever. In Search, Hydrogen is now the default model for AI Mode, and Google says it better understands long, multi-part questions and returns more visual answers with a wider range of links.
The honest caveat: nearly every number above comes from Google, so independent evaluations over the next few weeks will matter. Still, a fast model that plans less and finishes more is exactly what agentic workflows have been waiting for. Find more model launches in our AI news section.
Sources: Google: Introducing Gemini 4 Hydrogen — November 12, 2026; Google Developers: Build with Gemini 4 Hydrogen — November 12, 2026; 9to5Google — November 12, 2026.
More AI Stories

Gemini Google Wallet Integration: AI Budgeting Explained
Gemini can now read your Google Wallet spending to chart budgets, flag pricier subscriptions and plan holiday costs. Here's how the US rollout works.

XYBench: Why AI Assistants Miss the Real Question You Ask
Ai2's XYBench tests 8,115 misguided questions to see if AI assistants spot the real problem. The award-winning benchmark shows where models can improve.

SOV-Transcribe-1: Salesforce's Open Speech Model Explained
Salesforce's SOV-Transcribe-1 is an Apache 2.0 open-weight speech-to-text model with a 2.28% word error rate, style steering and 7 languages.
