AIFast Link gateway details

A third-party multi-model API relay built on New API. One key works with the OpenAI-compatible endpoint https://www.aifast.link/v1 (Chat Completions and Responses) and the Anthropic Messages endpoint https://www.aifast.link/v1/messages, and Gemini models also support the native Gemini API; the site configuration lists https://api.aifast.link as a backup address. On 2026-11-06 the public model catalog listed GPT (gpt-6-astra, gpt-6.1-sol, gpt-6.1-terra, gpt-5.6-sol, and others), Claude (claude-opus-5-5, claude-sonnet-5-5, claude-haiku-5-5, claude-fable-5-1, and others), and Gemini models (gemini-4-argon, gemini-3.8-flash, and others), plus the gpt-image-2 image model, with available models determined by the key's group. The developer docs include setup guides for Codex, Claude Code, Cursor, Dify, Cherry Studio, OpenWebUI, and other clients.

Best for: For developers outside mainland China who want one key for GPT, Claude, and Gemini and can pick the right key group for each client; start with a small pay-as-you-go top-up before deciding on a monthly plan.

Current status

Test yourself

Latency reference

Not tested

/models check

Public pricing API /api/pricing; /v1/models requires a key

Last verified

2026-11-06

Status notes

The website, model hub, developer docs, public pricing and site configuration APIs, user agreement, and privacy policy were accessible; keyless requests to /v1/models and /v1/messages both returned 401 authentication errors. Group names and account-pool descriptions are platform-reported, and plan prices and payment methods come from the docs' check snapshots of 2026-11-04 and 2026-10-08; AI Beyond has not signed up, made paid requests, or tested latency. The Uptime Kuma status page is disabled in the site configuration, and no independent public status page has been verified.

Pricing mode

Charged per token from an in-site USD credit balance, with the actual cost equal to the model's official USD price times the key group's multiplier; the dollar sign in the dashboard is an in-site usage unit, not cash. On 2026-11-06 the public pricing API showed Codex discount / dedicated / stable group multipliers of 0.7 / 1 / 1.1, Claude discount / stable groups at 0.9 / 1.5, Claude Code discount / stable / dedicated groups at 1.6 / 3.35 / 4.5, and the Gemini group at 1.8; group multipliers are the site's own billing coefficients, not official discounts. Pay-as-you-go: a 2026-10-08 check of the wallet page recorded in the docs shows CNY 7.50 paid for USD 10 of credit, and the docs give gpt-6.1-sol in the Codex discount group as an example, at a top-up-adjusted price of CNY 1.05 input and CNY 5.25 output per 1M tokens. There are also 30-day monthly plans with a daily quota that resets each day and covers only specified groups: in the docs' 2026-11-04 snapshot, GPT plans cost CNY 199 / 349 / 639 / 1199 for USD 20 / 40 / 80 / 160 of daily credit, and Claude plans cost CNY 258 / 449 / 889 / 1599 for the same four tiers; monthly plans do not include the Gemini group. The docs say the wallet page offers WeChat Pay and the monthly plan page offers Alipay and WeChat Pay, while credit cards and USDT are unconfirmed; check the dashboard and checkout page for current prices.

Price tags

Pay-as-you-go, Plan-based

Free quota

Public pages do not state a signup credit, and daily check-in is disabled

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://www.aifast.link/v1"
});

const response = await client.chat.completions.create({
  model: "gpt-6.1-sol",
  messages: [{ role: "user", content: "Test one call through an AI gateway." }]
});

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

• The user agreement (2026-10-08) prohibits using the service directly or indirectly within mainland China and prohibits providing its API capabilities to end users in mainland China through the platform; being able to open the site, sign up, or top up does not create an exemption, so users in mainland China should not use it. Not an official OpenAI, Anthropic, or Google service; official authorization and upstream sources have not been verified. Public group descriptions include "Claude official Max 20x account pool", "Kiro channel", "mixed account pool", "K12 channel", and "Antigravity + CLI", indicating that some routes come from account pools or third-party client access; AI Beyond has not verified their stability or compliance, and such routes may stop working at any time. Claude Code and Codex groups are meant for those clients and should not be treated as general endpoints. The privacy policy says request content is sent to upstream model providers and access channels; avoid sending sensitive data. The user agreement says unspent paid balance can be settled on request, while bonus benefits are not automatically convertible to cash; unused daily quota on monthly plans does not carry over, and auto-renewal and refund terms should be confirmed with support. The docs list WeChat aifast_club, QQ group 16595416, a Telegram group, and the website's support page as contact channels; the submitter also provided luwengwo@gmail.com, which is not shown publicly on the website. No independent public status page has been verified.
• Third-party relay pricing, models, and availability can change, so re-check them periodically.

Is AIFast Link 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 AIFast Link 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.