The best AI models from around the world

One key. One balance. Every model.

Claude, GPT, Gemini, DeepSeek, Kimi, Mistral and hundreds more through one OpenAI-compatible API. Route models your way, give every team its own model list, and keep spending under control.

Company accounts are set up by the GME AI team.

251
AI models, live today
32
leading AI labs
Worldwide
the strongest models from every leading lab
1
balance and one bill for all of them

Models

251 models, ready today

Claude, GPT, Gemini, DeepSeek, Kimi, Mistral and more from 32 leading AI labs. Your administrator chooses which ones your teams can use.

  • Google
  • OpenAI
  • Anthropic
  • Mistral AI
  • DeepSeek
  • Meta
  • NVIDIA
  • Moonshot AI
  • MiniMax
  • ByteDance Seed
  • Aion Labs
  • Amazon
  • Zhipu AI
  • Cohere
  • +18 more

Prices are shown in the console after you sign in.

Why A1 Token

AI for your whole business, without the hassle

One account gives every team the best AI models, with the reliability and control a business needs.

Connect once

One OpenAI-compatible API for every model. Switching models means changing one word, not rewriting code.

Always on

If a model is busy or down, the next one answers automatically, so your apps and customers keep going.

Spend less

Use a premium model only where it matters and a fast, low-cost one everywhere else. One balance, one bill.

Stay in control

Budgets for every key and team, a model list for every company, and a record of every call.

Route AI models

Your model line-up, one name to call

Create a route such as support-bot, line up the models behind it, and your apps call the route like any model. If the first model fails, the next one answers, and you pay once, for the model that served.

  • Automatic fallbackCombine models from different labs in one route, tried in your order.
  • Change without a releaseSwap a model in the console and every app on that route follows at once.
  • A default for everything elseSet the model that answers when a request asks for auto.

Customize AI models

Make the models fit your company

Every company and every team gets the models and names that suit it.

Your own model names

Give a route a name your team understands, then change the models behind it any time without touching code.

A model list for every team

Choose which models each company may use, and narrow it again for each key. Anything else is refused.

Your own default model

Choose the model that answers when an app asks for auto, and change it in the console without a release.

Built for companies

Everything your team needs, in one place

Stop juggling an account, a key and a bill for every AI vendor.

One API for every model

OpenAI-compatible. Change the base URL and keep your code, SDKs and tools. Switch models by changing one name.

Spending you control

A key for every person and service, each with its own limits and caps. When the balance reaches zero, calls stop.

Every call on the record

One ledger row per request: the model that ran, the key, the tokens and the cost. Written once, never edited.

Included with every account

  • Streaming and tool calls
  • A key per person or service
  • Rate limits per key
  • Daily and monthly spend caps
  • Company spend caps
  • Routes with fallback
  • A ledger of every call
  • Usage by key, model and day
  • Team roles and permissions
  • MCP for Cursor and Cline

How it works

Live in three steps

  1. 1
    Get your company accountThe GME AI team sets it up and adds your administrator.
  2. 2
    Create your keysOne for each person or service, each with its own limits and model list.
  3. 3
    Change one linePoint your OpenAI SDK at token.gme-ai.com/v1 and pick any model.

Python

from openai import OpenAI

client = OpenAI(
    base_url="https://token.gme-ai.com/v1",
    api_key=os.environ["A1_KEY"],
)

reply = client.chat.completions.create(
    model="claude-sonnet-5-5",  # or deepseek-v4-pro, kimi-k3, gpt-6.1-sol
    messages=[{"role": "user", "content": "Hello"}],
)

Node

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://token.gme-ai.com/v1",
  apiKey: process.env.A1_KEY,
});

const reply = await client.chat.completions.create({
  model: "deepseek-v4-pro",
  messages: [{ role: "user", content: "Hello" }],
});

cURL

curl https://token.gme-ai.com/v1/chat/completions \
  -H "Authorization: Bearer $A1_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"kimi-k3","messages":[{"role":"user","content":"Hello"}]}'

Questions

Good to know

Which models can I call?

251 models from Google, OpenAI, Anthropic, Mistral AI, DeepSeek, Meta, NVIDIA, Moonshot AI and 24 other labs, listed in the Models section. Use the name on each card. A name the gateway does not carry is refused with a 404 error, so you never pay for a model you did not ask for.

What is a route?

A name you create, such as support-bot, backed by an ordered list of models. Your apps call the route like a model; if the first model fails before it starts answering, the next one takes over, and you are billed once, for the model that answered.

Can each team use different models?

Yes. Your administrator chooses which models each company may use, and each key can be narrowed further. A request for a model outside that list is refused.

Do I need to rewrite my code?

No. A1 Token is OpenAI-compatible. Point the OpenAI SDK, or any tool that lets you set a base URL, at https://token.gme-ai.com/v1 and use your A1 key.

What happens when the balance runs out?

Calls stop with a 402 error. Nothing is charged to anyone else, and no call quietly switches to another account.

How do I get an account?

Company accounts are set up by the GME AI team. Your administrator then signs in, chooses the models and creates a key for each person or service. Prices are shown in the console.

The world's best AI models. One key to reach them.