Gemini 4 Hydrogen API
gemini-4-hydrogenGemini 4 Hydrogen is Google's fast Gemini 4 model for everyday agentic work, released on 12 November 2026. Google calls it the daily driver for agentic workflows: the model that writes code, runs tests and verifies changes, either on its own or as the worker under Gemini 4 Carbon or Argon, which Google keeps for deep architectural judgment and subtle bugs. Google also names it the recommended upgrade from Gemini 3.8 Flash. On APIMODELS it costs $0.825 per 1M input tokens and $4.125 per 1M output tokens, 45% below Google's $1.50 / $7.50 list, with cached input at $0.04125 (Google: $0.075). Google lists one rate for the whole 1M-token window, with no long-context surcharge and no scheduled price change. The Gemini 4 Hydrogen benchmarks Google published: 51 on the Artificial Analysis Intelligence Index at its default high thinking level, against 50 for Gemini 4 Argon, and 54 at xhigh; 57% on Terminal-Bench 4.1, against 18.2% for Gemini 3.8 Flash; 58.8% on HLE-Diamond; 71.6% on Agents' Last Exam v1.1; and a 96.7% honesty rate on Google's internal agent honesty evaluation. Google also says tasks finish 2.5 times faster than on Gemini 3.8 Flash at the default setting, and that early partners saw runs 2 to 2.8 times faster with 12% to 23% fewer tokens. These are Google's own figures, so run your own prompts before moving a production workload. Specifications: a 1,048,576-token context window shared by input, thinking and output (Gemini 3.8 Flash has a separate 64K output budget), up to 65,536 output tokens per response, text, image, video, audio and PDF input, and text output. Thinking levels are low, medium, high (the default) and xhigh; minimal is rejected with an error, and temperature, top_p and top_k are ignored, so drop them when you migrate from 3.8 Flash. Thinking tokens bill at the output rate, so estimate cost from total output rather than the visible answer, and lower the thinking level rather than squeezing max output tokens when you want to spend less. Caching is Google's implicit caching only, applied automatically to prompts of at least 4,096 tokens; whether a prefix hits is decided on Google's side, and explicit cache objects are not supported. Google lists function calling, structured outputs, code execution and grounding with Google Search for Hydrogen; we have not tested those through APIMODELS yet. Google offers a free tier with rate limits; here it is pay-as-you-go from one balance, with no Google Cloud project or overseas card needed. Call it with model gemini-4-hydrogen on the native Gemini endpoint or the OpenAI-compatible /v1/chat/completions, one API key for both; failed requests are not charged. When not to use it: for the hardest multi-step debugging and architecture calls, Google still points to Gemini 4 Carbon or Argon ($1.10 / $5.50 here through 30 November), and for very short, high-volume calls where price per token decides, Gemini 3.8 Flash ($0.450 / $2.250) costs less per token.
API Documentation
View complete API reference with all parameters and examples.
Full API Documentation
View complete API reference with streaming, thinking, and more.
Pricing
Billing: Cost = (input_tokens * input_price + output_tokens * output_price) / 1,000,000
45% Below Google List
$0.825 / $4.125 per 1M tokens (Google: $1.50 / $7.50); cached input $0.04125; one rate across the whole 1M window
Gemini 4 Hydrogen Benchmarks
Google: Artificial Analysis Intelligence Index 51 at default (Argon 50), 54 at xhigh; Terminal-Bench 4.1 57% (3.8 Flash 18.2%); HLE-Diamond 58.8%
Built to Execute
Google: the daily driver for agentic workflows — write code, run tests, verify changes — alone or as the worker under Carbon or Argon
Upgrade from 3.8 Flash
Google: tasks finish 2.5x faster at the default setting; early partners saw 12%–23% fewer tokens. Drop temperature / top_p / top_k and the minimal level
1M Shared Context, 65,536-Token Output
Input, thinking and output share 1,048,576 tokens; text, image, video, audio and PDF input; thinking levels low, medium, high (default), xhigh
Implicit Caching Only
Prompts of 4,096+ tokens are cached automatically on Google's side when they hit, at $0.04125 per 1M; no explicit cache objects
About Gemini 4 Hydrogen
Gemini 4 Hydrogen is a Large Language Model API provided by Google. Gemini 4 Hydrogen is Google's fast Gemini 4 model for everyday agentic work, released on 12 November 2026. Google calls it the daily driver for agentic workflows: the model that writes code, runs tests and verifies changes, either on its own or as the worker under Gemini 4 Carbon or Argon, which Google keeps for deep architectural judgment and subtle bugs. Google also names it the recommended upgrade from Gemini 3.8 Flash. On APIMODELS it costs $0.825 per 1M input tokens and $4.125 per 1M output tokens, 45% below Google's $1.50 / $7.50 list, with cached input at $0.04125 (Google: $0.075). Google lists one rate for the whole 1M-token window, with no long-context surcharge and no scheduled price change. The Gemini 4 Hydrogen benchmarks Google published: 51 on the Artificial Analysis Intelligence Index at its default high thinking level, against 50 for Gemini 4 Argon, and 54 at xhigh; 57% on Terminal-Bench 4.1, against 18.2% for Gemini 3.8 Flash; 58.8% on HLE-Diamond; 71.6% on Agents' Last Exam v1.1; and a 96.7% honesty rate on Google's internal agent honesty evaluation. Google also says tasks finish 2.5 times faster than on Gemini 3.8 Flash at the default setting, and that early partners saw runs 2 to 2.8 times faster with 12% to 23% fewer tokens. These are Google's own figures, so run your own prompts before moving a production workload. Specifications: a 1,048,576-token context window shared by input, thinking and output (Gemini 3.8 Flash has a separate 64K output budget), up to 65,536 output tokens per response, text, image, video, audio and PDF input, and text output. Thinking levels are low, medium, high (the default) and xhigh; minimal is rejected with an error, and temperature, top_p and top_k are ignored, so drop them when you migrate from 3.8 Flash. Thinking tokens bill at the output rate, so estimate cost from total output rather than the visible answer, and lower the thinking level rather than squeezing max output tokens when you want to spend less. Caching is Google's implicit caching only, applied automatically to prompts of at least 4,096 tokens; whether a prefix hits is decided on Google's side, and explicit cache objects are not supported. Google lists function calling, structured outputs, code execution and grounding with Google Search for Hydrogen; we have not tested those through APIMODELS yet. Google offers a free tier with rate limits; here it is pay-as-you-go from one balance, with no Google Cloud project or overseas card needed. Call it with model gemini-4-hydrogen on the native Gemini endpoint or the OpenAI-compatible /v1/chat/completions, one API key for both; failed requests are not charged. When not to use it: for the hardest multi-step debugging and architecture calls, Google still points to Gemini 4 Carbon or Argon ($1.10 / $5.50 here through 30 November), and for very short, high-volume calls where price per token decides, Gemini 3.8 Flash ($0.450 / $2.250) costs less per token. Through APIMODELS platform, you can access this model via a unified API with transparent pay-as-you-go pricing. Current pricing: Input: $0.825, Output: $4.125 per 1M tokens.
Key Features
- 45% Below Google List -- $0.825 / $4.125 per 1M tokens (Google: $1.50 / $7.50); cached input $0.04125; one rate across the whole 1M window
- Gemini 4 Hydrogen Benchmarks -- Google: Artificial Analysis Intelligence Index 51 at default (Argon 50), 54 at xhigh; Terminal-Bench 4.1 57% (3.8 Flash 18.2%); HLE-Diamond 58.8%
- Built to Execute -- Google: the daily driver for agentic workflows — write code, run tests, verify changes — alone or as the worker under Carbon or Argon
- Upgrade from 3.8 Flash -- Google: tasks finish 2.5x faster at the default setting; early partners saw 12%–23% fewer tokens. Drop temperature / top_p / top_k and the minimal level
- 1M Shared Context, 65,536-Token Output -- Input, thinking and output share 1,048,576 tokens; text, image, video, audio and PDF input; thinking levels low, medium, high (default), xhigh
- Implicit Caching Only -- Prompts of 4,096+ tokens are cached automatically on Google's side when they hit, at $0.04125 per 1M; no explicit cache objects
Use Cases
Chatbot & Customer Support
Build intelligent conversational systems to automatically answer user queries and improve service efficiency.
Content Generation
Automatically write articles, emails, ad copy, and other text content to boost productivity.
Code Assistant
Assist with code writing, debugging, and code review to accelerate software development.
Data Analysis
Understand and analyze unstructured data, extract key insights, and generate summary reports.
Why APIMODELS
- Unified API -- One API key to access all models, no need to register on multiple platforms
- Transparent Pricing -- Pay-as-you-go with per-call pricing published on every model page, ideal for indie developers and startups
- Instant Access -- Start right after signup; pay by card, PayPal, Alipay or USDT
- Full Documentation -- Detailed API docs with code examples in cURL, Python, and Node.js
Frequently Asked Questions
What is the difference between APIMODELS and the official Google API?
APIMODELS offers the same Gemini 4 Hydrogen model through our aggregation platform. We provide a unified API interface so you do not need separate accounts for each provider - one API key to access all models.
What is Gemini 4 Hydrogen API pricing on apimodels, and will it change?
$0.825 per 1M input tokens and $4.125 per 1M output, with cached input at $0.04125 — 45% below Google's list of $1.50 / $7.50 and $0.075 cached. Thinking tokens bill at the output rate, as they do at Google, and there is no long-context surcharge: Google lists one rate for the whole 1M window. Unlike Gemini 4 Argon and Carbon, Hydrogen has no scheduled price change on Google's side; if Google's list moves, this page will say so. You pay only for the tokens used, failed requests are not charged, and there is no subscription.
Gemini 4 Hydrogen vs Argon vs Carbon: which should I use?
Google splits the work this way: Hydrogen executes — writing code, running tests, verifying changes — and is the daily driver for agentic workflows, while Carbon and Argon handle the calls that need deep thought, such as designing a system or untangling a subtle bug. Google even suggests pairing them, with Carbon as the coordinator and Hydrogen as the worker and verifier. On capability, Google puts Hydrogen at 51 on the Artificial Analysis Intelligence Index at its default setting, just above Argon's 50. On price, Hydrogen is $0.825 / $4.125 per 1M tokens here, while Argon and Carbon are $1.10 / $5.50 through 30 November and double from 1 December along with Google's list. In practice, make Hydrogen the default for everyday coding and agent loops, and escalate to Carbon when a task stalls or a wrong answer is expensive. All three use the same key and endpoints; switching is the model field.
How do I migrate from Gemini 3.8 Flash to Gemini 4 Hydrogen, and is it worth it?
Google calls Hydrogen the recommended upgrade from 3.8 Flash, and the move takes four changes: set the model to gemini-4-hydrogen; remove temperature, top_p and top_k, which the server ignores; stop sending the minimal thinking level, which now returns an error (Hydrogen defaults to high and adds xhigh); and raise any hard-coded low maxOutputTokens, such as 8,192, to 65,536 — lower the thinking level instead when you want to spend less. The context budget changes too: 3.8 Flash has 1M of input plus a separate 64K of output, while Hydrogen shares 1,048,576 tokens across input, thinking and output. Is it worth it? Google reports Terminal-Bench 4.1 rising from 18.2% on 3.8 Flash to 57%, tasks finishing 2.5 times faster at the default setting, and early partners using 12% to 23% fewer tokens. Per token it costs more — $0.450 / $2.250 for 3.8 Flash here against $0.825 / $4.125 — so keep short, high-volume, price-driven calls on 3.8 Flash and move coding and multi-step agent work to Hydrogen.
Is Gemini 4 Hydrogen free?
At Google, within limits: Hydrogen is on the Gemini API free tier, with rate limits, in the countries and regions where Google offers the Gemini API. On APIMODELS it bills per token, and new accounts get $0.10 of trial credit (one new account per IP address every 48 hours). For scale, a request with 10,000 input tokens and 4,000 output tokens (thinking included) costs $0.00825 + $0.0165 = $0.02475, so the trial covers about four of them. After that you top up from $10 and pay only for what you use, with no subscription.
What are the Gemini 4 Hydrogen benchmarks?
From Google's launch post: 51 on the Artificial Analysis Intelligence Index at the default high thinking level, against 50 for Gemini 4 Argon, and 54 at xhigh; 57% on Terminal-Bench 4.1, against 18.2% for Gemini 3.8 Flash; 58.8% on HLE-Diamond; 71.6% on Agents' Last Exam v1.1; and a 96.7% honesty rate on Google's internal agent honesty evaluation. On speed, Google says tasks finish 2.5 times faster than on 3.8 Flash at the default setting. These are Google's own numbers rather than independent runs, so treat them as a starting point and compare on your own prompts — switching between Hydrogen, Argon, Carbon and 3.8 Flash here is a change to the model field.
How do I call the Gemini 4 Hydrogen API and set the thinking level?
Set the model to gemini-4-hydrogen and authenticate with your apimodels key. In the native Gemini format, POST https://api.apimodels.app/v1beta/models/gemini-4-hydrogen:generateContent (:streamGenerateContent for streaming) with a contents array, and set generationConfig.thinkingConfig.thinkingLevel to low, medium, high (the default) or xhigh. Google's advice is to start with high, keep xhigh for hard multi-step problems where high falls short, and drop to medium or low to spend less. In the OpenAI format, POST https://api.apimodels.app/v1/chat/completions, so the OpenAI SDK works once base_url is https://api.apimodels.app/v1. On launch day a one-word reply took us about 3 seconds; stream anything longer. You do not need a Google Cloud project or billing account, and it works from mainland China.
What are the Gemini 4 Hydrogen specs and limits?
A 1,048,576-token context window shared by input, thinking and output, and up to 65,536 output tokens per response; Google says longer outputs need background mode or long decode, which we have not verified through APIMODELS. Input can be text, image, video, audio or PDF; output is text only, with no image or audio generation and no Live API. Caching is implicit only: prompts of 4,096 tokens or more take part automatically and cached tokens bill at $0.04125 per 1M, but whether a prefix hits is decided on Google's side, so plan costs as if nothing is cached. Google lists function calling, structured outputs, code execution and grounding with Google Search; we have not tested those through APIMODELS yet, so try them before relying on them in production. Failed requests are not charged.
What can Gemini 4 Hydrogen do?
Gemini 4 Hydrogen supports: 45% Below Official, Fast Agentic Execution, 1M Context, Terminal-Bench 4.1 57%. See the APIMODELS docs for full parameters and call examples.
Can I use the Gemini 4 Hydrogen API from any country, including mainland China?
Yes. APIMODELS does not restrict access by country: the site and the API answer requests from anywhere, and apimodels.app and api.apimodels.app are reachable directly from mainland China without a proxy. You call Gemini 4 Hydrogen with one APIMODELS key, so you do not need a separate account with the model's developer.
What payment methods are supported? Can I pay with PayPal or crypto?
Yes to both. You can top up with a card (Visa, Mastercard and others, through Stripe), PayPal, Alipay, or USDT on TRON (TRC20) or BNB Smart Chain (BEP20). Card, PayPal and Alipay top-ups are credited immediately; USDT is credited once the transfer confirms on-chain. Auto top-up is optional, and a $50 top-up adds a $1 bonus ($3 on a $100 top-up). The balance is in USD and billed per use, with no subscription.