The Future of
Agentic Coding
Grok Code is a multi-agent programming system that builds, reasons, and evolves software autonomously.
[ CAPABILITIES ]
Models that fit your needs
Reasoning
Understand the universe: solve humanity's most difficult software problems with deep autonomous thought.
Vision
See the world through vision, interpreting interfaces and visual bugs with sharp, insightful understanding.
Tool Calling
Craft agentic experiences with both server-side and client-side tool calling with our Agent Tools API.
Multi-Agent Systems
Orchestrate crowds of specialized agents that collaborate, debate, and synthesize robust production code.
Code Generation
Bring your ideas to life with code generation, creating applications that are as unique as you are.
Real-time Search
Tap into the now with real-time search, pulling fresh, relevant documentation and context instantly.
Built for autonomous software creation
Unlike single-turn models, Grok Code is designed for complex, multi-step engineering tasks. It plans, acts, and iterates until the software is production-ready.
- multi-agent orchestration
- long-running execution
- tool integration
- self-improving loops
Models and pricing
| Model | Capabilities | Context window | Text Input | Image Input | Output |
|---|---|---|---|---|---|
Grok Code CoreNewOur latest flagship model with industry-leading speed and agentic tool calling capabilities. Combines the lowest hallucination rate on the market with strict prompt adherence. | 2M | $2.00 | $2.00 | $6.00 | |
Grok Code FastNewNon-reasoning variant of Grok Code Core for latency-sensitive use cases. | 2M | $2.00 | $2.00 | $6.00 | |
Grok Code Multi-AgentCost-efficient lightweight model that excels at tool-calling. | 2M | $0.20 | $0.20 | $0.50 |
Get started
API & SDK
Learn how to integrate Grok into your applications.
Console
Learn about billing, API keys, and account settings.
Grok App
Learn how to use Grok on web and mobile.