LinkAGI gateway details

A third-party multi-model API relay built on New API, aimed mainly at Chinese-speaking developers. Codex and OpenAI-compatible clients use https://api.linktoagi.com/v1 (Codex uses the Responses protocol), while Claude Code and Gemini CLI use https://api.linktoagi.com without /v1. The public model catalog lists GPT / Codex, Claude, Gemini, Grok, DeepSeek, and GLM models plus gpt-image image models, with available models determined by the key's group. The docs include setup guides for Codex, Claude Code, Gemini CLI, CC Switch, and Cherry Studio.

Best for: For Chinese-speaking developers using Codex, Claude Code, or Gemini CLI who can pick the right key group for each tool and are willing to test with a small balance first.

Current status

Test yourself

Latency reference

Depends on your deployment

/models check

Public pricing API /api/pricing; check the model hub for live models

Last verified

2026-11-05

Status notes

The website, setup docs, Codex guide, public pricing and status APIs, user agreement, and privacy notice were accessible. Group names and account-pool descriptions are platform-reported; AI Beyond has not signed up, made paid requests, or tested latency. The docs site's status and transparency page is a route check snapshot from 2026-08-04, not a continuously updated status page.

Pricing mode

Charged per token from a CNY balance, with the actual cost equal to the model's base billing times the key group's multiplier; image models are billed per request. On 2026-11-05 the public pricing API showed Codex Sale / Plus / Pro group multipliers of 0.10 / 0.15 / 0.25, Grok 0.2, DeepSeek 0.3, GLM 0.4, Gemini 0.6, and Claude groups ranging from 0.4 to 2. Group multipliers are the site's own billing coefficients, not USD-per-million-token prices or official discounts; check the model hub for unit prices. The docs list Alipay and WeChat Pay; the submitter says the wallet accepts custom top-ups from CNY 1.

Price tags

Pay-as-you-go

Free quota

A 2026-08-04 verification page on the docs site says email-verified signups receive CNY 0.2; a limited-time promotion also offers CNY 5 of trial credit for joining the QQ group and commenting on the V2EX promotion post, under the operator's posted rules

Usually you create an API key in the platform console, then replace the SDK's baseURL, model name, and auth details per the docs. Test streaming, function calling, error codes, and rate limits before launch.

import OpenAI from "openai";

const client = new OpenAI({
  apiKey: process.env.AI_GATEWAY_API_KEY,
  baseURL: "https://api.linktoagi.com/v1"
});

const response = await client.chat.completions.create({
  model: "gpt-6-sol",
  messages: [{ role: "user", content: "Test one call through an AI gateway." }]
});

• Supported models: OpenAI, Claude, Gemini, Grok, DeepSeek, GLM.
• Capabilities: OpenAI-compatible, Image generation.
• Documentation completeness: High; stability not tested.

• Not an official OpenAI, Anthropic, Google, or xAI service; official authorization and upstream sources have not been verified. Public group names include "Codex | Pro account pool (stable, no degraded output)", "Claude | Kiro reverse-engineered", "other platforms reverse-engineered", and "Max account pool - Claude Code only, no external clients", indicating that some routes come from account pools or reverse-engineered access; AI Beyond has not verified their stability or compliance, and such routes may stop working at any time. Codex groups restrict which clients can connect and should not be treated as a general OpenAI-compatible endpoint. The user agreement says unspent paid balance can be refunded, while bonus and promotional balance cannot; the verification page says invoices are available on request for top-ups above CNY 50. Usage logs and balance are shown in the authenticated dashboard; this does not establish a public balance-query or log-query API. The docs list the contact email linktoagi@163.com and QQ group 342205412, and the user agreement and privacy notice also list support@linktoagi.com. No continuously updated public status page has been verified.
• Third-party relay pricing, models, and availability can change, so re-check them periodically.

Is LinkAGI compatible with the OpenAI API?

This entry is marked as supporting the OpenAI-compatible format, but the exact base_url, model names, and parameters still follow the official docs.

Is LinkAGI ready for production?

Run a small-traffic load test, rate-limit test, and error-code test first, and confirm the billing rules; keep a backup provider for important workloads.