Skip to main content

google/gemini-4-carbon/chat

Google's Gemini 4 Carbon chat endpoint runs its strongest reasoning model on text and images, with function calling, JSON output and a 1,048,576-token shared context.

Input $2Output $10per 1 million tokensCached input $0.10

Workbench

Gemini 4 Carbon

What can Gemini 4 Carbon help with?

Missing the model you need?

DEEP REASONING CHAT API

Gemini 4 Carbon API at a glance

Google's strongest reasoning model, for problems that reward extra time spent checking the work.

Shared context
1M
1,048,576 tokens for input, thinking and reply
Reply ceiling
64K
65,536 tokens per call, thinking included
Gemini 4 tier
#1
Google calls Carbon its strongest model

Results

What Gemini 4 Carbon is good at

  • Positioning

    Reserved for the hard problems

    Google places Carbon above Gemini 4 Argon for multi-step debugging and open-ended research, while noting that Argon still suits most everyday engineering work.

  • Reasoning

    It spends tokens on checking itself

    Carbon uses more tokens to reason and verify than lighter models, so expect a long hidden working phase before a short final answer.

  • Long context

    One window for the whole problem

    Logs, source files and notes can share a single 1,048,576-token context with the reasoning and the reply.

Before you run it

What to know before you run Gemini 4 Carbon

Chat
Carbon reads text and images here. Convert audio, video and PDF files to text or page images before sending them.
Chat
Each reply stops at 65,536 tokens with thinking included. The long-decode mode that Google describes is not available on this endpoint.
Chat
Google's safety systems review every Carbon request, and some prompts may be declined.

Use cases

What to build with the Gemini 4 Carbon API

  • Chase a bug across several services

    Paste the failing logs and the relevant code, then ask Carbon to rank likely causes and name the check that would rule each one out.

  • Test a scientific argument

    Give Carbon a derivation or method section and ask where the reasoning depends on an assumption that has not been stated.

Prompts

Gemini 4 Carbon prompt examples you can copy

  • Chat

    Rank the likely causes

    These logs show an intermittent timeout between two services. Rank the three most likely causes and name one check that would confirm or rule out each.

    Why it works. Carbon does its best work when it must weigh several explanations and say how to test them.

    Try this prompt

Pricing

Gemini 4 Carbon API pricing

From $2 per million input tokens and $10 per million output tokens.

Get started

How to use the Gemini 4 Carbon API in three steps

  1. Step 1

    Leave plenty of room in max_tokens

    Thinking is drawn from the same budget as the answer. In our test, a 400-token cap on a long prompt spent 397 tokens on thinking and returned no text.

  2. Step 2

    Save Carbon for work that needs it

    Send routine rewrites and lookups to a lighter model, and bring Carbon in when a wrong answer would cost real time.

Parameter reference and code samples are in the API and Readme tabs at the top of this page.

FAQ

Gemini 4 Carbon API FAQ

About Gemini 4 Carbon

Reach for Gemini 4 Carbon when the problem is difficult enough to justify the extra reasoning: tangled bugs, careful technical review or research questions. Set a generous max_tokens value and state clearly what a finished answer looks like.

All Gemini models14

Text models9

Related models and next tasks