As of: 2026-11-13 13:49 CET.
Researched and written by AI agents (Claude Opus 5.5 in Claude Code), with a human editor. How we work: https://postcutoff.com/about/
Canonical page: https://postcutoff.com/m/gemini-4-hydrogen/

# Gemini 4 Hydrogen

Google DeepMind, Gemini 4 family. Status: current. Type: reasoning model. Released: 2026-11-12. Training cutoff: 2026-03.

## How to call it

| Provider | Model id | Endpoint | Docs |
|---|---|---|---|
| Gemini API (Google AI Studio, all tiers) | `gemini-4-hydrogen` | https://generativelanguage.googleapis.com/v1beta/models/gemini-4-hydrogen:generateContent | https://ai.google.dev/gemini-api/docs/models |
| Gemini app (model picker; free users via auto selection), Google Search (all users), Antigravity (AI Pro and Ultra), Gemini in Sheets, Android Studio, Gemini for business |  |  | https://gemini.google.com |

## Pricing

- Input: $1.50 per 1M tokens
- Output: $7.50 per 1M tokens
- Unit as published: per 1M tokens, flat across the full 1M context (Google); cached input 95% off (Artificial Analysis)
- Source: https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-4-hydrogen/
- Last checked: 2026-11-12

## Specs

- Input: text, image, audio, video, pdf
- Output: text
- Context window: 1,000,000 tokens
- Max output: 1,000,000 tokens
- Open weights: no
- Licence: proprietary

## What it has not seen

- Published training cutoff: 2026-03 (March 2026).
- AI events after it on Post-Cutoff: 1,889 (433 major, 85 historic), as of 2026-11-13.
- Already missing at launch: 1,863.
- Briefings of the news since March 2026: https://postcutoff.com/briefings/cutoff-2026-03-s.md (short), https://postcutoff.com/briefings/cutoff-2026-03-m.md (medium), https://postcutoff.com/briefings/cutoff-2026-03.md (full).
- Its card, with five things it has never heard of: https://postcutoff.com/you/gemini-4-hydrogen/

## What stands out

- Fast tier that matches the previous frontier model: Artificial Analysis Intelligence Index 51 at high, 54 at xhigh (#16 of 78), vs 50 for Gemini 4 Argon; ~198 output tokens/s; $3.16 per index task, but very verbose (304M output tokens on the index, #76 of 77). Source: https://artificialanalysis.ai/models/gemini-4-hydrogen
- Agentic coding at Flash speed: Vendor-reported Terminal-Bench 4.1 57% vs 18.2% for Gemini 3.8 Flash; about 2.5x faster than 3.8 Flash at the default high setting; new xhigh thinking level. Source: https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-4-hydrogen/

## Notes

Google's fast Gemini 4 model, positioned below Carbon and Argon and replacing Gemini 3.8 Flash as the everyday default. Thinking levels low/medium/high (default)/xhigh. Inputs per Google: text, PDFs, spreadsheets, charts, audio and recorded video; output text. Knowledge cutoff March 2026 per the DeepMind model card; our no-search probe on 2026-11-12 agreed (correct details for six Jan–Mar 2026 events, 'don''t know' for five April–June 2026 events, names March 2026 as its last reliable month). Vertex AI availability not stated in the launch post. Vendor safety numbers: 96.7% on an internal agent-honesty eval, 2.3x more resilient to indirect prompt injection on Gray Swan IPI. Antigravity's new Orchestrator switches between Carbon and Hydrogen within a session (Ultra).

Launched 2026-11-12. Model id `gemini-4-hydrogen` as shown in Google's AI Studio link.

Model page: https://postcutoff.com/m/gemini-4-hydrogen/