Gemini 4 Hydrogen
Google's fast Gemini 4 model for agentic workflows and everyday coding, and its recommended upgrade from Gemini 3.8 Flash.
- Previous version: Gemini 3.8 FlashSee what changed
Google's fast Gemini 4 model for agentic workflows and everyday coding, and its recommended upgrade from Gemini 3.8 Flash.
Overview
Gemini 4 Hydrogen is built for quick, multi-step agent work: calling tools, writing and fixing code, and working across text, images, video, audio and PDFs. It thinks at four adjustable levels, defaulting to high, and through the Gemini API it can search the web, run code, read URLs and operate a computer.
Google recommends it as the upgrade from Gemini 3.8 Flash and says it finishes tasks 2.5 times faster at its default setting. Google also reports a slightly higher Artificial Analysis Intelligence Index score than Gemini 4 Argon, 51 against 50; both are provider claims, not Cody measurements.
Category comparison
These are provider-published specifications, not Cody benchmark scores. Follow the linked sources for current limits and endpoint-specific exceptions.
Pricing & comparisons
Set your usage. Your estimate updates as you type.
Example: 6,000 input tokens for 10 pages, plus a 500-token summary. Page lengths vary; adjust the numbers below.
One run sends your input to the model once and receives an answer. Tokens are pieces of text: input is what you send, output is the answer you receive.
Gemini 4 Hydrogen
Estimated total (USD)
For the usage above · USD · API pricing, not a subscription
Estimates exclude taxes, tools, cache storage/writes, free allowances and custom discounts. Image estimates cover output only, not prompt or reference-image charges. Quality modes differ by model. Unlisted settings are not treated as free.
API and provider access
Availability
Available through the Gemini API and Google AI Studio on free and paid tiers, and in the Gemini app, Google Search and Google Antigravity.
Access is limited to Google's supported regions and eligible accounts.
Check live availabilityData and training
On the Gemini API's free tier, Google may use prompts and responses to improve its products; on the paid tier, it does not.
This is a concise reading of the cited provider material, not legal advice. A third-party gateway can have different storage, routing, training, and residency terms from the model maker's direct API.
Read the provider policyFrequently asked questions
Google's fast Gemini 4 model for agentic workflows and everyday coding, and its recommended upgrade from Gemini 3.8 Flash. Gemini 4 Hydrogen is built for quick, multi-step agent work: calling tools, writing and fixing code, and working across text, images, video, audio and PDFs. It thinks at four adjustable levels, defaulting to high, and through the Gemini API it can search the web, run code, read URLs and operate a computer.
Gemini 4 Hydrogen was released on November 12, 2026 according to the cited provider materials.
Available through the Gemini API and Google AI Studio on free and paid tiers, and in the Gemini app, Google Search and Google Antigravity. The access routes listed in this guide are Google.
$1.50 input · $7.50 output / 1M tokens. Rates stay flat across the full 1M-token context window, and thinking is billed as output. Cached input costs $0.075 per 1M tokens through implicit caching. Batch and Flex cost half; Priority costs $2.70 input and $13.50 output. A rate-limited free tier is also available.
Available through the Gemini API and Google AI Studio on free and paid tiers, and in the Gemini app, Google Search and Google Antigravity. Access is limited to Google's supported regions and eligible accounts.
On the Gemini API's free tier, Google may use prompts and responses to improve its products; on the paid tier, it does not. The policy belongs to the provider route and account terms, so verify it again before production use.
One price per model, using the same input/output mix. We weight each API rate by its share of one million total tokens and use the lowest matching reviewed route. This is a comparison rate, not a cost per task or a quality score. Tokenizers, caching, reasoning and long context can change your actual bill.
75% input + 25% output · no cache discounts
Anthropic
OpenAI
Mistral AI
OpenAI
OpenAI
Anthropic
This model
OpenAI
Anthropic
OpenAI
Anthropic
OpenAI
Meta
OpenAI
Anthropic
OpenAI
Anthropic
Anthropic
Anthropic
OpenAI
Anthropic
OpenAI
Related comparisons
Google's most capable Gemini 4 reasoning model, built for complex software engineering and mathematical and scientific problem solving.
Open full comparisonAn OpenAI model for interpreting images, working with application screens and producing clear written responses.
Open full comparisonGoogle's Gemini 4 model for coding, research and tasks involving large collections of text, images, audio or video.
Open full comparisonA Claude model for everyday coding, writing and visual analysis, with room for large reference documents.
Open full comparisonGoogle's stable, high-efficiency multimodal model for agents, software work, and large mixed-media inputs at an introductory Flash-tier price.
Open full comparison