Gemini 4 Hydrogen API

Updated 14 Nov 2026

What is the Gemini 4 Hydrogen API?

The Gemini 4 Hydrogen API is live on UnificAlly as google/gemini-4-hydrogen, from $0.45/1M tokens of input and $2.25/1M tokens of output. Gemini 4 Hydrogen is Google's fast Gemini 4 model for everyday coding and agent work, released November 12, 2026. It is the third model in the Gemini 4 family, after Gemini 4 Argon and Gemini 4 Carbon, and the first with an xhigh reasoning level. Input and output share one 1M-token window, and one response returns up to 64K tokens. It runs on UnificAlly as google/gemini-4-hydrogen, with text and image input and four reasoning effort levels (low, medium, high, xhigh).

Gemini 4 Hydrogen API at a glance

The Gemini 4 Hydrogen API bills per token, with no subscription. Thinking tokens are billed as output tokens. The Gemini API pricing guide covers the rates for every Gemini model.

How to use the Gemini 4 Hydrogen API

1. Create an API key on the API keys page. The same key and balance work for every model on UnificAlly.
2. Send a POST request to /v1/chat/completions on https://direct-api.unifically.com with the model ID and a messages array. This is the OpenAI-compatible endpoint shared by every language model on UnificAlly.
3. Read the reply from choices[0].message.content in the same response, or from the stream when stream is true. There is no task to check.

curl -X POST https://direct-api.unifically.com/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -d '{
    "model": "google/gemini-4-hydrogen",
    "reasoning_effort": "high",
    "stream": true,
    "messages": [
      {"role": "user", "content": "Find the bug in this function, explain it in two sentences, and write the fix."}
    ]
  }'

Tokens arrive as they generate. The stream is sent as server-sent events and ends with data: [DONE]. Lower reasoning_effort to low for quick steps, and raise it to xhigh for the hardest ones. A reply at xhigh can take a minute or more to start, so stream it and give your client timeout some headroom. Send requests to direct-api.unifically.com so long replies are not cut short by a proxy timeout.

Without stream, the response comes back synchronously with the completion and the token usage:

{
  "id": "3f8e2a1c-7b4d-4e9a-9c2f-5d1e8b7a6c40",
  "object": "chat.completion",
  "created": 1710000000,
  "model": "google/gemini-4-hydrogen",
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "The loop stops one item early, so the last element is never checked..."
      },
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 28,
    "completion_tokens": 64,
    "total_tokens": 92,
    "prompt_tokens_details": {
      "cached_tokens": 0
    },
    "completion_tokens_details": {
      "reasoning_tokens": 0
    }
  }
}

Gemini 4 Hydrogen API parameters

The Gemini 4 Hydrogen API documentation has the full reference.

Image input request

Send an image in an image_url content part alongside your text, either as a direct https link or inline as a base64 data URI. The link must point straight at the image file, with no redirect. Images are counted as input tokens.

curl -X POST https://direct-api.unifically.com/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -d '{
    "model": "google/gemini-4-hydrogen",
    "messages": [
      {"role": "user", "content": [
        {"type": "text", "text": "Read every number in this chart and return them as a JSON table."},
        {"type": "image_url", "image_url": {"url": "https://example.com/chart.png"}}
      ]}
    ]
  }'

Gemini 4 Hydrogen key features

What Gemini 4 Hydrogen is best for

• Everyday coding agents

  57% on Terminal-Bench 4.1 and 87.8% on Vibe Code Bench v1.1 at a speed that keeps long loops moving.

• High-volume agent steps

  About 170 output tokens per second, with low and medium effort for the cheap steps.

• Tax and finance analysis

  #1 on Tax Agent Bench at 52.3% and #2 on the Vals Finance Index at 66.1%.

• Hard reasoning on demand

  54 on the Artificial Analysis Intelligence Index when a step needs xhigh.

• Whole-codebase context

  A 1M-token window with up to 64K tokens out in one response.

• Charts and screenshots

  Image input for reading diagrams, tables, and UI screenshots alongside text.

Gemini 4 Hydrogen use cases

Make it the default model in a coding agent. It reads the repo, runs the tests, and fixes the failure without the wait of a larger model. Run worker and verifier agents on it while a stronger model plans the job. Build tax and finance pipelines that read filings and check the numbers against each other. Send it a chart or a screenshot and ask for the data as JSON. Keep low or medium effort for classification and extraction, and raise a single step to xhigh when it gets stuck.

Gemini 4 Hydrogen limitations

xhigh is slow and it writes a lot. Artificial Analysis measures about 43 seconds before the first answer token at xhigh, against 18 seconds at high. It used 304M output tokens to finish the Intelligence Index at xhigh and 110M at high. Thinking tokens bill as output, so the top level costs far more per task than the per-token rate suggests.

It is not the strongest Gemini 4 model. Gemini 4 Carbon scores 60 on the Artificial Analysis Intelligence Index, 70.5% on the Vals Index, and 63.1% on Terminal-Bench 4.0, against 54, 68.5%, and 54.0% here.

Images go in as direct https links or inline base64 data URIs, and a link that redirects is refused. Video and audio input are not available on this model yet, and one response tops out at 64K tokens.

Gemini 4 Hydrogen vs Gemini 4 Carbon

Carbon is the stronger reasoner and Hydrogen is the faster worker. Carbon leads the Artificial Analysis Intelligence Index 60 to 54, the Vals Index 70.5% to 68.5%, and Terminal-Bench 4.0 63.1% to 54.0%. Hydrogen writes about 170 tokens per second against about 80 for Carbon, starts answering sooner, and edges Carbon on Tax Agent Bench, 52.3% to 51.8%. Hydrogen also has the xhigh level that Carbon lacks.

Gemini 4 Hydrogen vs Gemini 3.8 Flash

Hydrogen replaces Gemini 3.8 Flash for agent work. It finishes tasks 2.5x faster at the default setting, and Terminal-Bench 4.1 goes from 18.2% to 57%. Both share the 1M-token window. Move agent loops over first, since that is where the gap is widest.

When to use Gemini 4 Hydrogen

Use Gemini 4 Hydrogen as the everyday model: coding agents, tool-calling loops, extraction, and document analysis where speed counts. Step up to Gemini 4 Carbon when the task is long, the answer has to be right, and waiting is fine. Try xhigh on a hard step before you switch models.

Gemini 4 Hydrogen guides and comparisons

Review pricing, limits, and tested outputs before running this model.

Gemini 4 Hydrogen API pricing

Gemini API pricing per 1M tokens for Gemini 4 Argon and 3.8 Flash, Gemini TTS per second, Google's upcoming price changes, and what real requests cost.

Read Gemini 4 Hydrogen API pricing→