GPT Image 2.5 is live — OpenAI's newest image model, targeted edits that leave the rest of the frame alone

Gemini 4 Hydrogen — Fast Coding and Agent Execution

Use Gemini 4 Hydrogen for everyday code changes, interactive apps and tool-driven work. Read mixed documents in a shared 1M context, with four thinking levels for different tasks.

Multimodal inputFunction callingStructured outputStreaming
Highlights
1M shared context · four thinking levels
Best for
Everyday coding, interactive apps and tool-using agents
Input
Text, images, video, audio and PDF
Output
Text · up to 65,536 tokens per hop
Gemini 4 Hydrogen

Ask Gemini 4 Hydrogen

Start with the prompt in the composer or choose an example.

Enter to send · Shift+Enter for a new line

Pricing

Credit-based — 1 credit = $0.001 USD. Pay only for completed generations.

Pricing by group · gemini-4-hydrogen

Loading groups…

USD per 1M tokens. Tool charges may apply.

What is Gemini 4 Hydrogen?

Gemini 4 Hydrogen is Google's stable model for everyday coding and agent execution, released on November 12, 2026. Gemini 4 Hydrogen reads text, images, video, audio and PDF and produces text. Input, thinking and output share one 1,048,576-token context window. Google positions Gemini 4 Hydrogen for responsive coding loops, interactive applications and parallel worker tasks; Carbon is aimed at harder reasoning problems, while Argon handles advanced long-horizon work. Try Gemini 4 Hydrogen on a task whose result you can check, such as a patch with a failing test or a document extraction with a known schema.

What you can build with this model

Real-world workflows and production use cases you can build and ship with this model.

Gemini 4 Hydrogen coding workflow concept with an editor panel, patch and test tile.

Ship the next fix with Gemini 4 Hydrogen

Give Gemini 4 Hydrogen the failing test, relevant code and the behavior you expect. Ask for a focused patch and a regression test, then run the test in your own environment. For interactive UI work, include a screenshot and describe the specific change. Keep repository instructions and acceptance criteria in the request so Gemini 4 Hydrogen has enough context to work within your existing conventions.

Gemini 4 Hydrogen agent workflow concept with separate task cards and a review step.

Coordinate small tasks with Gemini 4 Hydrogen

Use Gemini 4 Hydrogen as a worker for bounded jobs: inspect one module, propose tests for one function, or check a proposed fix against a requirement. Your application can call tools and combine the results. Google's suggested multi-model workflow uses Carbon for difficult planning and Gemini 4 Hydrogen for execution. Keep permissions and final decisions in your application; a function call from the model is a request for your code to act.

Gemini 4 Hydrogen multimodal reading concept with documents, an image and an audio waveform.

Read across files with Gemini 4 Hydrogen

Combine a PDF, a screenshot or an audio recording with a question that names the evidence you need. Gemini 4 Hydrogen can extract fields, explain a chart or compare statements across inputs, returning text or a requested structured format. Ask Gemini 4 Hydrogen for source locations and check them. A large context window lets you supply more material, but it does not make an unsupported answer reliable.

Fit Gemini 4 Hydrogen to the task

Gemini 4 Hydrogen thinking levels

Gemini 4 Hydrogen defaults to high and also offers low, medium and xhigh. Google recommends evaluating the levels on your own tasks and choosing the lowest one that meets your quality bar. Reserve xhigh for difficult work where the extra thinking helps; it can use more tokens and take longer.

Gemini 4 Hydrogen native requests

Gemini 4 Hydrogen uses the GenerateContent format for complete JSON responses or incremental streaming. Native responses preserve function calls, thought signatures and continuation state. Applications can keep the same request structure when comparing Gemini 4 Hydrogen with Carbon or Argon, while setting the thinking level deliberately.

Give Gemini 4 Hydrogen the evidence

Give Gemini 4 Hydrogen the code, document passages and visual details needed for one decision. The Gemini 4 Hydrogen context supports work across those inputs. Ask it to identify missing information and cite the material behind its answer, then check the result against your sources.

Try Gemini 4 Hydrogen on your work

  1. step 01

    Choose a Gemini 4 Hydrogen group

    Open the Gemini 4 Hydrogen Playground with your existing reAPI account. Compare the current group prices and choose one. Start with a real bug, a small UI change or a document question whose result you can check.

    Open
  2. step 02

    Compare Gemini 4 Hydrogen thinking levels

    Start with high for Gemini 4 Hydrogen, then compare low, medium and xhigh on the same task. Google recommends keeping the lowest level that meets your quality bar. Record total token use as well as answer quality.

    Open
  3. step 03

    Verify Gemini 4 Hydrogen results

    Run generated tests, validate extracted fields and inspect proposed tool arguments. Gemini 4 Hydrogen can still make mistakes. Keep consequential actions behind your existing permission checks and review process.

    Open

Capabilities

Gemini 4 Hydrogen multimodal input

Text, images, video, audio and PDF are documented inputs. Responses are text. Image generation, audio generation and the Live API are outside this model's supported output surface.

Gemini 4 Hydrogen shared context

Gemini 4 Hydrogen shares 1,048,576 tokens across input, thinking, function calls and final output. Leave room for the answer when sending a large prompt; these are not separate additive allowances.

Gemini 4 Hydrogen thinking controls

Gemini 4 Hydrogen offers low, medium, high and xhigh, with high as the default. Thinking cannot be switched off. Available thought summaries are distinct from the final answer and may not be present in every response.

Gemini 4 Hydrogen tools and output

Google documents function calling, code execution, grounding, file search and structured output for Gemini 4 Hydrogen. Your code executes client functions. Check tool availability for your request and validate structured results against the schema you supplied.

Gemini 4 Hydrogen streaming

Gemini 4 Hydrogen permits up to 65,536 output tokens per hop in both JSON and SSE. Longer generation uses continuation within the remaining shared context. Preserve the original request and returned token when continuing.

Gemini 4 Hydrogen implicit caching

Gemini 4 Hydrogen supports implicit caching for eligible prompts, with a documented minimum of 4,096 tokens. Explicit cached-content resources are unsupported. Returned usage identifies cache hits; repeating a prompt does not guarantee one.

Gemini 4 Hydrogen vs Gemini 4 Carbon

Google gives these models different roles: Gemini 4 Hydrogen for rapid execution, Carbon for deeper reasoning. The comparison below uses documented behavior, not a reAPI performance benchmark. Argon is another option for advanced long-horizon tasks.

Decision point
Gemini 4 Hydrogen
Gemini 4 Carbon
Suggested work
Everyday coding, interactive apps and worker agents
Hard architecture, debugging, science and security reasoning
Thinking levels
low / medium / high / xhigh
low / medium / high
Default thinking
high
medium
Unified context
1,048,576 tokens, shared
1,048,576 tokens, shared
Non-streaming REST hop
Up to 65,536 output tokens
Up to 32,768 output tokens
Streaming REST hop
Up to 65,536 output tokens
Up to 65,536 output tokens

Sources: Google's model documentation and Hydrogen migration guide, checked November 14, 2026. Actual task quality, latency and total cost depend on the request. These are documented specifications, not measured reAPI performance.

Frequently asked questions

Common questions about this model.

Google positions Gemini 4 Hydrogen for everyday coding, interactive applications, tool-using agents and multimodal document workflows. Gemini 4 Hydrogen can also act as a worker or verifier in a larger agent system. Evaluate it with tasks and acceptance checks that match your application.

docs/api/gemini-4-hydrogen

API reference

Drop-in code and the full parameter table.

curl https://reapi.ai/api/v1beta/models/gemini-4-hydrogen:generateContent \
  -H "x-goog-api-key: YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"contents":[{"role":"user","parts":[{"text":"Fix a Python mutable default argument and add a regression test."}]}],"generationConfig":{"maxOutputTokens":4096,"thinkingConfig":{"thinkingLevel":"high"}}}'

Parameter validation failed

Rejected synchronously with the failing field named. Check enum values, ranges, and URL fields — nothing is charged.

Authentication or balance issue

401 means a missing or invalid Bearer key; 402 means the reserve exceeds your balance. Manage keys and credits in the console.

Content or material rejected

Reference material and outputs pass automated moderation. Rejected tasks fail with a clear error and a full refund.

Task failed or timed out

A task that reaches a failed state is never charged — the reserve refunds automatically. Keep the task id and retry when ready.

start building

Try Gemini 4 Hydrogen with a checkable task

Choose a Gemini 4 Hydrogen group, compare its current rates and send a native request with your existing reAPI key. Use the request guide to understand thinking levels, streaming and continuation.