Google · LLMs & Coding

Gemini 4 Carbon

Catalog data · checked Nov 16, 2026

Gemini 4 Carbon is a language model from Google, released in November 2026. It reads text, images, video, audio and PDF files and writes text, with a context window of 1.05M tokens and up to 32.8K output tokens per answer. It costs $2 per million input tokens and $10 per million output tokens. That puts it in the upper price range: 77% of the 185 paid language models we track cost less.

Calculate my costCompare

Input price$2per 1M tokens
Output price$10per 1M tokens
Context window1.05Mtokens per request
Max output32.8Ktokens per answer
ReleasedNov 20268 days ago
Knowledge cutoffMar 2026training data ends
WeightsClosedAPI only

What it can do

  • Tool calling
  • Structured output
  • Reasoning mode
  • Image input
  • File input
  • Audio input
  • Video input
  • Image output
  • Audio output
  • Prompt caching

More prices

Cached input (read)$0.1 /1M
Audio input$2 /1M

What it costs in practice

1,000 support replies1,000 chats with 800 input and 250 output tokens each$4.1
Summarize a 300-page reportabout 150,000 input tokens in one request, 2,000 tokens out$0.32
10,000 product descriptions300 input and 200 output tokens per text$26
One day of agentic codingabout 3M input and 150K output tokens across many calls$7.5

Calculated from the list prices above. Your bill depends on your real token counts. Use your own numbers →

Compare Gemini 4 Carbon with

Questions & answers

How much does Gemini 4 Carbon cost?

Gemini 4 Carbon costs $2 per 1 million input tokens and $10 per 1 million output tokens (USD, list price checked Nov 16, 2026).

How large is the context window of Gemini 4 Carbon?

Gemini 4 Carbon can process up to 1,048,576 tokens per request and return up to 32,768 tokens. As a rule of thumb, 1,000 tokens are about 750 English words.

Does Gemini 4 Carbon support tool calling?

Yes. The API accepts tool (function) definitions and also supports structured JSON output.

Can Gemini 4 Carbon read images?

Yes, image input is supported, as well as file input (for example PDFs).