Gemini 4 Hydrogen
Google's fast, efficient Gemini 4 model for everyday coding and agentic workflows. Accepts text and an optional image plus a system prompt and reasoning effort, billed per token.
Gemini 4 Hydrogen is Google's fast, efficient model in the Gemini 4 series, adapted from Gemini 4 Carbon and built for everyday coding, agentic workflows, and rapid execution. It supports a 1M-token context window and four thinking levels, and succeeds Gemini 3.8 Flash.Muapi exposes Gemini 4 Hydrogen through one simple endpoint with text and image input, an optional system prompt, a reasoning_effort control (low, medium, high, xhigh), token-based billing at Google's official $1.50 per million input tokens and $7.50 per million output tokens, and the same asynchronous submit-and-poll contract as every other model.
Gemini 4 Hydrogen is available now via API
Try Gemini 4 Hydrogen in the playground, or call it from your own code with a Muapi API key. See the Gemini 4 hub for Gemini 4 Argon, Gemini 4 Carbon, and every other Gemini 4 model.
Google's fast, efficient Gemini 4 model for everyday coding and agentic workflows. Accepts text and an optional image plus a system prompt and reasoning effort, billed per token.
Gemini 4 Hydrogen is the fast model in Google's Gemini 4 series. Google describes it as a fast, efficient model for everyday coding and agentic workflows, adapted from Gemini 4 Carbon and succeeding Gemini 3.8 Flash. It supports a context window of up to 1M tokens (input and output combined) and accepts text, images, video, audio, and PDFs as input. Benchmark figures Google reports for it are Google's own results, not Muapi measurements.
Muapi wraps it as a single gemini-4-hydrogen endpoint. Send a prompt with an optional image_url, system_prompt, and reasoning_effort (low, medium, high, or xhigh), then poll the shared result endpoint. For deeper frontier reasoning, see Gemini 4 Carbon or Gemini 4 Argon.
Built for quick turnaround on everyday coding, question answering, and summarization.
Pass long documents or large parts of a codebase in one request, with a context window of up to 1M tokens as reported by Google.
Set reasoning_effort to low, medium, high (default), or xhigh to trade speed and cost for depth.
Include an image alongside your prompt to ask about screenshots, diagrams, or scanned documents.
Pay $1.50 per million input tokens and $7.50 per million output tokens, with no subscription.
Write, fix, and refactor code with focused edits and fast turnaround.
Answer questions over long inputs and summarize reports, transcripts, and contracts.
Use Hydrogen as a fast reasoning core for multi-step agents and automated workflows.
Serve chat and support assistants where responsiveness and per-token cost matter.
Ask questions about screenshots, diagrams, and scanned pages by attaching an image.
| Feature | Gemini 4 Hydrogen | Gemini 4 Carbon | Gemini 3.8 Flash |
|---|---|---|---|
| Positioning | Fast, efficient everyday coding | Security-focused frontier reasoning | Fast, low-cost chat |
| Input | Text, image, system prompt, reasoning effort | Text, image, system prompt, reasoning effort | Text, image, system prompt |
| Reasoning effort levels | low, medium, high, xhigh | low, medium, high | Not configurable |
| Price (input / output per 1M tokens) | $1.50 / $7.50 | $4.00 / $20.00 | $2.50 / $12.50 |
| Minimum charge per call | $0.00015 | $0.0004 | $0.0002 |
| Best for | Everyday coding and agentic workflows | Vulnerability analysis and secure code | Real-time assistants and high-volume tasks |
Submit a JSON request with your Muapi API key, save the request ID, and poll the shared result endpoint for the completed text.
curl -X POST https://api.muapi.ai/api/v1/gemini-4-hydrogen \
-H "x-api-key: YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{"prompt":"Refactor this function to remove the nested loops and explain the change","system_prompt":"You are a senior software engineer. Be concise.","reasoning_effort":"high"}'
# Response: {"request_id":"REQUEST_ID"}Call gemini-4-hydrogen with a prompt, and optionally a system_prompt, image_url, and reasoning_effort.
curl https://api.muapi.ai/api/v1/predictions/REQUEST_ID/result \ -H "x-api-key: YOUR_API_KEY" # Poll until status is completed, then read the generated text from outputs.
Poll until the task is completed, then read the generated text from outputs.
Gemini 4 Hydrogen is Google's fast, efficient model in the Gemini 4 series, adapted from Gemini 4 Carbon and built for everyday coding and rapid execution. It succeeds Gemini 3.8 Flash.
Billing is per token at $1.50 per million input tokens and $7.50 per million output tokens, with a minimum charge of $0.00015 per call. Reasoning tokens count as output. There is no subscription.
It sets how much the model reasons before answering: low, medium, high (default), or xhigh. Higher effort improves depth on hard problems but is slower and uses more output tokens.
Yes. Pass an image_url with your prompt to ask questions about a screenshot, diagram, or photographed document.
The endpoint is asynchronous. You receive a request_id immediately, then poll the predictions result endpoint or provide a webhook URL to be notified when the text is ready.
Hydrogen is the fast, lower-cost model for everyday coding and agentic work, while Carbon is the security-focused frontier tier. Both use the same prompt, image_url, system_prompt, and reasoning_effort fields; Hydrogen adds an xhigh level.
No. Any benchmark figures Google reports for Gemini 4 Hydrogen are Google's own results. Run your own evaluations on your workload before relying on the model.
Add fast, efficient Gemini 4 reasoning to your product with one API.