grok2api

Product Introduction

1. Definition: Grok2API is an open-source, self-hosted API gateway and management platform built in Go with a React admin console. It is a technical solution designed to unify and streamline programmatic access to multiple xAI Grok AI services.
2. Core Value Proposition: It exists to solve the problem of fragmented authentication and complex operational management for developers and teams using multiple Grok AI accounts across different services (Grok Build, Grok Web, Grok Console) simultaneously. Its core value is providing a single, unified OpenAI- and Anthropic-compatible API endpoint.

Main Features

1. Multi-Provider Account Pool Management: The gateway maintains independent, isolated pools of accounts for Grok Build, Grok Web, and Grok Console. It handles credential storage (with encryption), automatic quota synchronization, and model capability discovery for each account. The system supports bulk import/export via JSON/JSONL/TXT and provides tools for credential renewal and account cleanup.
2. Intelligent Request Routing & Failover: Features a dynamic routing layer that selects the optimal account based on the requested model, available quota, and health status. It supports model discovery (especially for Grok Build), provider pinning, sticky sessions for prompt cache affinity, and bounded failover within a selected provider to maintain consistency. Concurrency and quota guards prevent account overuse.
3. Unified Multi-Protocol API Layer: Exposes a consolidated API that is compatible with both OpenAI and Anthropic SDKs. It supports /v1/chat/completions, /v1/responses (for Grok Build's native protocol), and Anthropic's /v1/messages endpoint. The gateway handles protocol translation, session management, response compaction, and stores conversation history for retrieval.
4. Advanced Egress & Proxy Management: Includes a sophisticated egress manager with scoped proxy pools (HTTP, SOCKS, Resin) for each provider (Build, Web, Console). It features connection failure probes with immediate recovery checks, proxy-pool leasing to avoid global cooldowns, and integration with FlareSolverr for Cloudflare clearance. The optional Egress Quality Guard sidecar performs active health checks using real model calls.
5. Comprehensive Admin Operations Dashboard: A built-in React admin console provides full visibility and control. Admins can manage client API keys with rate/spend limits, view real-time model routes and account status, audit usage logs, configure runtime settings, and manage generated media (images, videos) with local archiving.

Problems Solved

1. Pain Point: Developers and organizations face significant overhead managing authentication tokens (OAuth, SSO), rate limits, and quota for multiple Grok AI accounts across different platforms (Build, Web, Console). Manually switching between accounts and credentials is inefficient and error-prone.
2. Target Audience: The primary users are AI Developers and Engineering Teams building applications on Grok's APIs, DevOps Engineers responsible for deploying and maintaining AI service infrastructure, and Organizations requiring stable, load-balanced, and auditable access to Grok AI models at scale.
3. Use Cases: Essential for building resilient AI applications that need fallback across multiple accounts; for teams sharing a pool of paid Grok accounts; for developers who require a single endpoint compatible with popular AI SDKs (OpenAI, Anthropic); and for scenarios requiring detailed audit logs for API usage and billing reconciliation.

Unique Advantages

1. Differentiation: Unlike simple proxy wrappers, Grok2API is a full-fledged gateway with account lifecycle management, intelligent routing, and a professional admin UI. It goes beyond basic API forwarding by handling provider-specific protocols (like Grok Build's Responses), model discovery, and structured failover logic, which DIY scripts or generic API managers lack.
2. Key Innovation: Its provider-scoped, intelligent egress system is a key innovation. By isolating proxy pools per service and implementing immediate, bounded failure probes, it maximizes connection reliability without imposing long global cooldowns. The Egress Quality Guard extends this by using actual AI model calls for health checks, enabling proactive quarantine of unreliable proxies.

Frequently Asked Questions (FAQ)

1. What is Grok2API and how does it work with Grok AI? Grok2API is a self-hosted API gateway that consolidates access to xAI's Grok Build, Grok Web, and Grok Console services. It works by pooling your accounts from these services, managing their credentials and quotas, and presenting them through a single, unified API endpoint that is compatible with OpenAI and Anthropic client libraries.
2. How does Grok2API handle account authentication and security? The gateway securely stores account credentials (like OAuth refresh tokens and SSO cookies) encrypted in its database using a configurable key. It handles the automatic renewal of Grok Build OAuth tokens and can manage SSO sessions for Grok Web/Console, isolating these sensitive details from your application code.
3. Can I use Grok2API to load balance requests across multiple Grok accounts? Yes, load balancing and failover are core features. You can pool multiple accounts for a single model (e.g., several Grok Build accounts with grok-4.5). The gateway's routing logic will distribute requests based on quota availability and health, providing automatic failover if one account is rate-limited or encounters an error.
4. What are the deployment requirements for running Grok2API? For a single instance, you can run it with Docker Compose using the provided image, which requires minimal setup (just a config.yaml file). For high-availability multi-instance deployment, you need a shared PostgreSQL database, a Redis instance for runtime storage, and a shared read-write directory for media files.
5. Does Grok2API support image and video generation models from Grok? Yes, it fully supports Grok's image generation (grok-imagine-image), image editing, and video generation models (grok-imagine-video). The gateway handles the API calls to Grok Web or Console for these tasks and can optionally archive the generated media files locally for later retrieval.