{"slug": "google-announces-gemini-4-hydrogen-rolling-out-to-gemini-app", "title": "Google announces Gemini 4 Hydrogen, rolling out to Gemini app", "summary": "Google unveiled Gemini 4 Hydrogen, a model it says surpasses Gemini 4 Argon on the Artificial Analysis Intelligence Index at its default high thinking level (51 vs. 50, reaching 54 at its highest xhigh setting), and is rolling it out to the Gemini app as the \"Everyday help\" option. Google said Hydrogen finishes tasks 2.5 times faster at its high default setting than Gemini 3.8 Flash, more than triples Terminal-Bench 4.1 scores (57% vs. 18.2%), cuts fabricated claims by more than 5x, and supports native 1 million-token input and up to 1 million output tokens. Google also said partners saw Hydrogen run 2x to 2.8x faster while reducing token usage by 12% to 23% versus 3.8 Flash, and that it reduces overreach in agentic tasks by up to 81% and cuts misaligned actions and task shortcutting by 12x.", "body_md": "After this week’s [Carbon availability](https://9to5google.com/2026/11/10/gemini-app-4-carbon/), Google today unveiled Gemini 4 Hydrogen instead of keeping the “Flash” branding. \n\n### Benchmarks\n\nGoogle says it brings “cutting-edge intelligence to everyone — with speed and efficiency for everyday coding and agentic workflows.” Notably:\n\n“Hydrogen surpasses Argon on the [Artificial Analysis Intelligence Index](https://artificialanalysis.ai/) at its default high thinking level (51 vs. 50, and reaching 54 at its highest xhigh thinking setting), while often completing tasks faster than models in its class.”\n\nHydrogen supports native 1 million-token input and up to 1 million output tokens.\n\nThis follows the launch of [Gemini 3.8 Flash](https://9to5google.com/2026/09/02/gemini-3-8-flash-launch/) on September 2, with Google afterwards focusing on frontier models with Argon and Carbon. Compared to 3.8 Flash, Hydrogen is billed as a “is a “major standalone upgrade” with “significant leaps in both speed and intelligence.” Specifically:\n\n- Hydrogen finishes “tasks 2.5 times faster at its high default setting”\n- “…more than [triples] Terminal-Bench 4.1 scores (57% vs. 18.2%)”\n- Cuts “fabricated claims by more than 5x”\n- Adapts “cleanly when a tool call fails or a search returns empty”\n- “…our partners have seen Hydrogen run 2x to 2.8x faster while reducing token usage by 12% to 23%, compared to 3.8 Flash.”\n- “On multi-step tasks like deep research, data analysis, and broad codebase navigation, Hydrogen ranks #5 on [Agent Arena](https://arena.ai/leaderboard/agent) , arena.ai’s live benchmark.”\n- “For deep domain reasoning and analysis, it leads its class on HLE-Diamond (58.8%), ranks #1 overall on [Vals Tax Agent Bench](https://www.vals.ai/benchmarks/tax_agent_bench) (52.3%), ties for #1 on the[Vals Legal Research Bench](https://www.vals.ai/benchmarks/legal_research) (55.8%), and leads its class on[AutomationBench-AA](https://artificialanalysis.ai/evaluations/automationbench-aa) .”\n- “It outperforms other models on the new [Multimodal Knowledge Agent (MKA)](https://blog.google/innovation-and-ai/models-and-research/gemini-models/multimodal-knowledge-agent-benchmark/) benchmark (53.4%) for end-to-end agentic knowledge work over complex visual files, and on[Agents’ Last Exam v1.1](https://agents-last-exam.org/leaderboard) (71.6%) for end-to-end professional workflows.”\n- Compared to 3.8 Flash, 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 faced with impossible or blocked tasks, it is up to 2x less likely to bypass permission boundaries and achieves a 96.7% honesty rate on our internal agent honesty evaluation.”\n\n### Gemini app\n\nIn the Gemini app, Hydrogen is better at following and sticking to your tone, format, and rules, while offering clear tables and visual breakdowns. As such, “email drafts, study guides, rewrites, and documents come back ready to use on the first try.” Additionally, Hydrogen “more frequently calls the right tools (i.e. Connected Apps) and skills for the job.”\n\nIn the model picker, 4 Hydrogen is billed as providing “Everyday help.”\n\n- **Quick answers** : 3.5 Flash-Lite\n- **Everyday help** : 4 Hydrogen\n- **Detailed research** : 3.1 Pro (on AI Pro)\n- **Code & analysis** : 4 Argon\n- **Most complex work** : 4 Carbon\n\nFree users in the Gemini app get access to Hydrogen via the “auto model selection when it’s the appropriate model for the prompt.”\n\n### Google Search\n\nGoogle says Hydrogen significantly upgrades Search (AI Mode) by better understanding your intent:\n\n…you can ask your question in your natural flow of thought — even if you have multiple parts to your query, hard-to-understand typos, or highly specific requests — and Search will intelligently comprehend what you’re looking for.\n\nResponses get more visual and will feature a wider range of relevant links.\n\n### Antigravity\n\nIn Google Antigravity, 4 Hydrogen has an “Extra” thinking tier. Then there’s Orchestrator where “Carbon that may offload routine steps to Hydrogen. Same Carbon-level quality, fewer tokens used.”\n\nTo help make this multi-model collaboration automatic, we’re introducing [Orchestrator in Google Antigravity](https://antigravity.google/blog/gemini-4-hydrogen-in-google-antigravity). Rather than using a single model from start to finish, Orchestrator dynamically switches between models like Carbon and Hydrogen within the same session.\n\nFor workflows that require parallel agents, Carbon can act as the coordinator while Hydrogen can power worker and verifier agents to build and test changes simultaneously across isolated workspaces — helping teams ship complex projects faster without driving up token costs.\n\nFor developers, Hydrogen costs $1.50 per million input tokens and $7.50 per million output tokens.\n\nGemini 4 Hydrogen is available “for all tiers of [Google AI Studio](https://aistudio.google.com/prompts/new_chat?model=gemini-4-hydrogen), the Gemini API, [Gemini for business](https://cloud.google.com/gemini-enterprise), the [Gemini app](http://gemini.google.com/), [Google Search](https://blog.google/products-and-platforms/products/search/gemini-4-hydrogen-search-upgrade), and paid Google AI Pro and Ultra subscribers in [Google Antigravity](https://antigravity.google/).”\n\n*Updating…*\n\n*FTC: We use income earning auto affiliate links.* [More.](https://9to5mac.com/about/#affiliate)\n\n[our homepage](http://9to5google.com/)for all the latest news, and follow 9to5Google on\n\n[exclusive stories](https://9to5google.com/feature/exclusive/),\n\n[reviews](https://9to5google.com/guides/reviews/),\n\n[how-tos](https://9to5google.com/guides/how-to/), and\n\n[subscribe to our YouTube channel](https://www.youtube.com/9to5google)", "url": "https://wpnews.pro/news/google-announces-gemini-4-hydrogen-rolling-out-to-gemini-app", "canonical_source": "https://9to5google.com/2026/11/12/gemini-4-hydrogen-launch/", "published_at": "2026-11-12 18:01:41+00:00", "updated_at": "2026-11-12 18:23:12.164695+00:00", "lang": "en", "topics": ["artificial-intelligence", "large-language-models", "generative-ai", "ai-agents", "ai-products"], "entities": ["Google", "Gemini 4 Hydrogen", "Gemini 4 Argon", "Gemini 4 Carbon", "Gemini 3.8 Flash", "Gemini app", "Artificial Analysis Intelligence Index", "Terminal-Bench 4.1"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/google-announces-gemini-4-hydrogen-rolling-out-to-gemini-app", "markdown": "https://wpnews.pro/news/google-announces-gemini-4-hydrogen-rolling-out-to-gemini-app.md", "text": "https://wpnews.pro/news/google-announces-gemini-4-hydrogen-rolling-out-to-gemini-app.txt", "jsonld": "https://wpnews.pro/news/google-announces-gemini-4-hydrogen-rolling-out-to-gemini-app.jsonld"}}