AI
Overview
About this model
Gemini 4 Carbon is Google's reasoning model for complex software engineering, scientific problem solving, cybersecurity analysis, and sustained agent workflows. It accepts text, images, audio, video, and PDFs and generates text. Input, thinking, and output share a 1,048,576-token context. Standard synchronous generation supports up to 65,536 thinking and output tokens; longer generation requires Google's long-decode background or continuation mode and remains within the shared context limit. It supports low, medium, and high thinking levels, tool calling, structured output, and implicit context caching.
- Context window
- 1048576
- Category
- -
- Last updated
- 2026-11-12
- Providers
- -
Performance
Usage and availability
Last 14 days
Requests over timeDaily volume
Availability over timeDaily success rate
Cache hit rateDaily cache reads
API endpoints
Supported parameters
Availability
Providers
-
| Provider | Route | Pricing | Cache hit | TTFT | Output speed (TPS) | Availability |
|---|
Provider performanceAll active provider routes
API access
Request example
Choose a protocol and copy a ready-to-run request.
POST/v1/responses