Multi-provider pool · Auto failover · Direct access

One key, every frontier model.

LLM API is a multi-model API relay for developers: Claude, GPT, Gemini, DeepSeek, GLM, Qwen and Kimi behind one key, natively speaking the Anthropic, OpenAI and Gemini protocols. Subscription or metered billing, direct access, set up in two environment variables.

Claude Code / Anthropic API OpenAI SDK / Codex CLI Gemini API / SDK
Get Free API Key Read the Docs
Full Tool Support No VPN Required Unlimited Tokens
Configure Claude Code in One Click

# one command, fully configured

PS> irm llmapi.pro/setup.ps1 | iex

# or two env vars, manually

$ export ANTHROPIC_BASE_URL=https://llmapi.pro

$ export ANTHROPIC_API_KEY=your-key

$ claude

Works with all Claude Code terminals and IDE plugins

Claude◆ GPT◆ Gemini◆ DeepSeek◆ GLM◆ Qwen◆ Kimi◆ MiniMax◆ Grok◆ Anthropic Protocol◆ OpenAI SDK◆ Codex CLI◆ Gemini Protocol◆ Claude◆ GPT◆ Gemini◆ DeepSeek◆ GLM◆ Qwen◆ Kimi◆ MiniMax◆ Grok◆ Anthropic Protocol◆ OpenAI SDK◆ Codex CLI◆ Gemini Protocol◆

01 — Models

Every Major Model, One Relay

Flagship models from 10+ vendors, channels and pricing synced live

Anthropic Claude

Anthropic Claude

Flagship coding & reasoning · powers Claude Code

claude-opus-4-8 · claude-sonnet-5
OpenAI GPT

OpenAI GPT

All-round · Codex coding agents

gpt-5.6 · codex
Google Gemini

Google Gemini

Multimodal · long context

gemini-3.1-pro · flash
DeepSeek

DeepSeek

Best value

GLM

GLM

Domestic flagship

Qwen

Qwen

Tongyi lineup

Kimi

Kimi

Long context

MiniMax

MiniMax

Fast multimodal

Grok

Realtime

Image

Per request

Browse all channels & live pricing

02 — Billing

Subscribe monthly, or pay as you go

Two independent billing modes — run both on one account.

Subscription

Heavy coding · capped cost

¥59from / mo
  • Unlimited tokens at a fixed monthly price
  • Full Claude / GPT / Gemini / Grok chat lineup on one quota
  • Exclusive features like Remote Session
View plans

Metered API

Pay per use · every model

Top up & go
  • Balance never expires — no monthly fee
  • Every model on the platform — domestic, image gen, multimodal
  • Transparent pricing: channels, multipliers and rates published live
Browse pricing Open API Console

03 — The Problem

The multi-model pain

Official APIs: pricey and scattered

One account, one invoice and one SDK per vendor — and per-token bills that add up fast.

Pro plan runs out in 45 min

Rate limits kill your flow. You hit the cap right when you need it most.

Alternatives are a nightmare

LiteLLM breaks, proxies fail, tool calling doesn't work. Hours wasted debugging.

We solved all three.

04 — How It Works

Up and running in 60 seconds

Three steps. No configuration files. No proxy servers.

1

Sign up

Get your API key in 30 seconds. Free tier included.

2

Set two variables

Add these to your shell and you're done.

3

Code as usual

Claude Code works exactly the same. No changes needed.

terminal

# Point your client at LLM API

$ export ANTHROPIC_BASE_URL=https://llmapi.pro

$ export ANTHROPIC_API_KEY=your-key

# That's it. Start coding.

$ claude

05 — Included

Everything you need

Built for Claude Code, Codex CLI and any SDK.

Three protocols, natively

Anthropic / OpenAI / Gemini protocol compatible: messages, streaming, tool calling, extended thinking — any client connects with zero changes.

Remote Session (Exclusive)

Control your Claude Code from phone — scan QR, chat syncs, seamless cross-device. Not available on official Claude.

Usage Dashboard

Track tokens, monitor costs, manage API keys. All in one place.

Always On

Multi-provider failover. If one backend goes down, we switch automatically. Your work never stops.

06 — Why Us

Why LLM API

Goodbye sky-high token bills

Official API charges per token — one feature can cost tens of yuan. Our fixed monthly fee, unlimited tokens, one price no matter how much you use. Plans start at ¥59/mo, while the equivalent official experience runs into the hundreds. Making AI coding affordable.

Production-grade reliability

Multi-node load balancing, automatic failover, smart 429 retry. Your workflow won't be interrupted by backend fluctuations. 24/7 uptime to support your development rhythm.

Direct connect, zero setup

No VPN, no credit card, no overseas phone number. Two env vars and you're ready. Alipay payment in CNY. Solving the biggest barrier for Chinese developers using frontier models.

07 — Billing

Two billing modes, your choice

Subscriptions with unlimited tokens; metered API with a balance that never expires.

Trial

¥1

One-time trial · 40 requests

Get Started

Pro

¥59/mo

Unlimited · 2,500 req/week

Get Started
Recommended

Max

¥249/mo

Unlimited · 10,000 req/week

Get Started

Ultra

¥979/mo

Unlimited · 40,000 req/week

Get Started
See full plans and feature comparison → Live metered API pricing →

08 — Voices

What developers say

”

Switched from the official API and saved $150 in the first month. Setup took literally two minutes. Everything just works.

M

Marcus T.

Full-stack Developer

”

I was burning through my Pro plan in under an hour. Now I code all day without worrying about limits. The tool support is flawless.

S

Sarah K.

Backend Engineer

”

Our team of 4 moved to the Team plan. We went from $800/month combined to $99. The failover means we've had zero downtime.

J

James R.

Engineering Lead

09 — FAQ

FAQ

What kind of service is LLM API?
LLM API is a multi-model API relay. It exposes Anthropic-, OpenAI- and Gemini-compatible endpoints so tools like Claude Code CLI and Codex CLI (which support a configurable backend via documented environment variables) can route requests through a multi-provider pool with automatic failover and unified billing. LLM API is independent and not affiliated with any model vendor.
Will my tools work?
Yes. Read, Write, Edit, Bash, Grep, Glob — everything works. LLM API conforms to the Anthropic API tool-calling specification, so Claude Code CLI behaves the same way as with any other compatible backend.
Why use LLM API instead of going direct?
LLM API is a relay service that routes requests across multiple compatible upstream providers, with automatic failover and unified billing. You get higher availability and a single integration point in exchange for a small relay overhead.
Can I switch back to going direct anytime?
Anytime. Just remove the <strong>ANTHROPIC_BASE_URL</strong> environment variable and Claude Code CLI will connect to its default endpoint again. No lock-in, ever.
Is my code safe?
We don't store your code. Requests are proxied in real-time and not logged. Your intellectual property stays yours.

Start coding for free

No credit card required. Set up in under a minute.

Get Free API Key