Google

Gemini 4 Carbon

Google's most capable reasoning model, built to think deeply and check its work

Pro plan requiredTextPowerful

Available on the Pro plan.

Input
$2 per 1M tokens
Output
$10 per 1M tokens
Reads up to
1.0M tokens
Writes up to
66K tokens

Google’s list price. Your own use is counted in credits.

About

Google’s most capable reasoning model, for complex software engineering, maths and science, and multi-step agents. Spends more thinking per turn than Argon. 1M context.

Capabilities

  • Vision
  • Audio
  • Web search
  • Thinking
  • Functions

Best for

  • Complex coding
  • Math & science
  • Multi-step agents
  • Factual accuracy

Compare Gemini 4 Carbon

All comparisons

More from Google

All Google models