Skip to content
NewA calmer, faster ZeroHour
StoryModel releaseinfo

Google rolls out Gemini 4 Hydrogen across Search and AI products

7 sources · 8 articlesFirst reported Updated ()
What's new

November 13: A MarkTechPost capture repeats the 2.5-times speed claim and adds no specifications, pricing, or access details.

Merged summary · x-ai/grok-4.7 · rewritten as coverage arrivesSummaries by Vani AI

Gemini 4 Hydrogen is rolling out globally in Search and other Google products, replacing Gemini 3.8 Flash.

Google is rolling out Gemini 4 Hydrogen globally in Search for all users and in the Gemini app, AI Studio, the API, business plans, and Antigravity. It replaces Gemini 3.8 Flash and, unlike Argon and Carbon, is not limited to the costliest subscription. Google says it is the new default in AI Mode and is used for the hardest AI Overviews, with low, medium, high, and a new xhigh thinking level. The company claims high mode is 2.5 times faster than 3.8 Flash, fabricated claims fall more than fivefold, and Terminal-Bench rises from 18.2% to 57%. Ultra subscribers can use Orchestrator, which Google says nears Carbon quality, uses 24% less quota, and finishes 19% faster.

Key facts
  • Google cites Artificial Analysis scores of 51, and 54 at the highest setting.
  • Simon Willison lists Hydrogen at $1.50 per million input tokens and $7.50 output.
  • Carbon and Argon introductory rates double on December 1, 2026, Willison reports.
  • Orchestrator switches between Hydrogen and Carbon for Ultra and business users.
  • Pro and Ultra get Hydrogen in Antigravity; Search access is for all users.
Explain this AI news with Vani AIWhat was announced, what is new and why it matters, in a new chat

Timeline

dated events in the reporting · oldest first
  1. Nov 30, 2026Introductory Carbon and Argon token rates end · Simon Willison
  2. Dec 1, 2026Carbon and Argon token rates are set to double · Simon Willison

What each outlet adds

beyond the shared facts
  • GoogleSearch availability for all users, query fan-out, and AI Mode languages
  • Hacker NewsOrchestrator quota and speed claims, plus Terminal-Bench and index scores
  • Ars TechnicaNot limited to the costliest plan; fabricated-claims claim versus Flash
  • The VergeDefault in AI Mode and claimed parity with named rival models
  • Simon WillisonHydrogen, Carbon, and Argon token prices and the December increase

Coverage

oldest first · one row per article · UTC
  1. · 6d ago Google · AIFirst report

    Google Search rolls out Gemini 4 Hydrogen to all users

    Google says the model improves multi-part questions, query fan-out, visual answers, and responses in every supported AI Mode language.

  2. · 6d ago Hacker News · AI

    Google rolls out Gemini 4 Hydrogen and Orchestrator

    Google says Orchestrator nears Carbon quality while using 24% less quota and finishing 19% faster.

  3. · 5d ago Ars Technica · AI

    Google releases Gemini 4 Hydrogen, replacing Gemini 3.8 Flash

    Google says high mode is 2.5 times faster than 3.8 Flash and cuts fabricated claims more than fivefold.

  4. · 5d ago TechCrunch · AI

    Google Search adds Gemini 4 Hydrogen for complex questions

    Google says Hydrogen is 2.5 times faster than Gemini 3.8 Flash and is rolling out to Search users globally.

  5. · 5d ago The Verge · AI

    Google releases Gemini 4 Hydrogen as the default Search AI Mode model

    It is live across Search, the Gemini app, and the API, with a new xhigh thinking level.

  6. · 5d ago Simon Willison

    llm-gemini 0.37 notes Gemini 4 Hydrogen launch and prices

    Hydrogen is the cheapest Gemini 4 model at $1.50 per million input tokens; Carbon and Argon prices double December 1.

  7. · 5d ago MarkTechPost

    Google releases Gemini 4 Hydrogen, title says 2.5x faster than Gemini 3.8 Flash

    The captured page is only a cookie-consent banner, so no specs, availability, or benchmarks are in the text.

  8. · 54m ago TechCrunch · AI

    Google adds temporary threads to AI Mode in Search

    Temporary threads detach AI Mode chats from the account so replies ignore past preferences and do not shape future personalization.