gemini-4-hydrogen

Gemini 4 Hydrogen API — price per token

Google’s fast Gemini 4 model for everyday coding and agentic workflows, with four reasoning levels, incremental streaming and a shared 1M-token context.

Pricing per 1M tokens

RateOfficial GoogleHere (−50%)
Input$1.5$0.75
Cached input$0.075$0.0375
Output$7.5$3.75

Input

Official Google
$1.5
Here (−50%)
$0.75

Cached input

Official Google
$0.075
Here (−50%)
$0.0375

Output

Official Google
$7.5
Here (−50%)
$3.75

Every request is metered at the official rate first, then the flat 50% B2C discount is subtracted before it touches your prepaid balance. Context window: 1M tokens shared by input, thinking and output. Max output: 64K streaming · 32K non-streaming.

Best for

  • High-volume coding, agents and chat where speed and price matter most.
  • Function-calling agents and JSON schema output.
  • Documents, images, audio and video in one shared context.

Good to know

  • Use gemini-4-hydrogen. Reasoning levels are low, medium, high and xhigh; high is the default. minimal and thinkingBudget are unsupported. Sampling controls are ignored.
  • Text, inline image, WAV audio, MP4 video and PDF input; text output. Native Gemini, Chat Completions, Responses and Messages share one key and balance.
  • Input, thinking and output share 1,048,576 tokens. This route supports up to 65,536 streaming or 32,768 non-streaming output tokens; long-decode continuation is unavailable.
  • Implicit caching bills cached input at 5% of the input rate. Explicit caches, Batch and server-side tools are unavailable on this model. The final user turn must contain text or only tool results.
  • Official rates are flat across the 1M context: $1.50 input, $0.075 cached input and $7.50 output per 1M tokens, thinking included. The account discount applies after the tariff.

How to use Gemini 4 Hydrogen

Create a free account, generate one key, and point any Gemini-compatible tool at https://router.apitoken.sale with model ID gemini-4-hydrogen — the native /v1beta/models/gemini-4-hydrogen:generateContent surface works, authenticated with x-goog-api-key. Top up your balance before sending API requests. Existing integrations on the legacy host https://gemini.api.apitoken.sale keep working.

Frequently asked questions

Which model ID do I use?

Use gemini-4-hydrogen on router.apitoken.sale. Select low, medium, high or xhigh with thinkingLevel in Gemini requests or reasoning_effort in Chat Completions.

How is Hydrogen different from Carbon?

Hydrogen is the fast Gemini 4 model for everyday coding and agents, with an extra xhigh reasoning level and high as its default. Carbon is Google’s strongest Gemini 4 model for the hardest tasks. Both share the same context and output limits.

What was verified?

Incremental streaming, all four reasoning levels, JSON schema, function calls and tool results, implicit caching, image/audio/video/PDF input and long-context input. Batch, server-side tools and long-decode continuation are unavailable.

Run Gemini 4 Hydrogen on the native Gemini API at a flat 50% off — instant key, prepaid balance, card or crypto.