Setup guide

Set up your API in 60 seconds

Every tool below needs the same three things: our endpoint URL, your API key and a model ID. Pick your tool, paste, done. Endpoint: https://api.unlimitedai.sh/v1

You only need three things

  • →Endpoint URL: https://api.unlimitedai.sh/v1
  • →API key: the sk-... key you received after your purchase
  • →Model ID: any ID from the models page, e.g. z-ai/glm-5.3

Your settings, in plain text

If your tool asks for these fields, fill them in exactly like this. That is the whole setup for most tools.

Universal settings
Endpoint URL: https://api.unlimitedai.sh/v1
API Format: Chat Completions (/chat/completions)
API Key: your sk-... key
Model ID: z-ai/glm-5.3

Set up ZCode (recommended)

ZCode lets you save the key once and reuse it in every chat. Open Settings → Model Settings, then:

  • →Click Add Provider and give it a name, e.g. unlimitedai
  • →Endpoint URL: paste https://api.unlimitedai.sh/v1
  • →API Format: leave it on Chat Completions (/chat/completions)
  • →API Key: paste your sk-... key
  • →Under Model List, click Add Model and type a model ID such as z-ai/glm-5.3 — repeat for every model you want
  • →Toggle the provider to Enabled and pick one of its models in chat
ZCode Model Settings screen showing how to add a custom API provider: endpoint URL, API format, API key and model list
ZCode → Settings → Model Settings → Add Provider: enter the endpoint URL, API key and your models

Claude Code

Paste this one-liner in your terminal, replacing sk-xx with your key.

One-liner
ANTHROPIC_BASE_URL='https://api.unlimitedai.sh' ANTHROPIC_AUTH_TOKEN='sk-xx' CLAUDE_CODE_DISABLE_NONESSENTIAL_TRAFFIC=1 claude --model 'anthropic/claude-opus-4.8'

Codex CLI

One-liner
UAI_API_KEY='sk-xx' codex -m 'openai/gpt-6-luna' -c 'model_provider="uai"' -c 'model_providers.uai={name="unlimitedai.sh",base_url="https://api.unlimitedai.sh/v1",env_key="UAI_API_KEY",wire_api="responses"}'

OpenCode

One-liner
UAI_API_KEY='sk-xx' OPENCODE_CONFIG_CONTENT='{"provider":{"uai":{"npm":"@ai-sdk/openai-compatible","name":"unlimitedai.sh","options":{"baseURL":"https://api.unlimitedai.sh/v1","apiKey":"{env:UAI_API_KEY}"},"models":{"z-ai/glm-5.3":{}}}}}' opencode -m 'uai/z-ai/glm-5.3'

Aider

One-liner
OPENAI_API_BASE='https://api.unlimitedai.sh/v1' OPENAI_API_KEY='sk-xx' aider --model 'openai/anthropic/claude-sonnet-5'

Cursor

In Cursor: Settings → Models → enable Override OpenAI Base URL, then fill in:

Settings
Override OpenAI Base URL: https://api.unlimitedai.sh/v1
OpenAI API Key: sk-xx
Custom model: z-ai/glm-5.3

Cline, Roo Code and Windsurf

All three use the same extension settings screen. Choose OpenAI Compatible as the provider and fill in:

Extension settings
API Provider: OpenAI Compatible
Base URL: https://api.unlimitedai.sh/v1
API Key: sk-xx
Model ID: anthropic/claude-sonnet-5

Continue

Open ~/.continue/config.yaml and add this block, then reload the editor.

config.yaml
models:
  - name: unlimitedai GLM
    provider: openai
    model: z-ai/glm-5.3
    apiBase: https://api.unlimitedai.sh/v1
    apiKey: sk-xx

VS Code (Copilot custom models)

settings.json
"github.copilot.chat.customOAIModels": {
  "z-ai/glm-5.3": { "name": "GLM 5.3", "url": "https://api.unlimitedai.sh/v1", "toolCalling": true, "maxInputTokens": 128000, "maxOutputTokens": 16000, "requiresAPIKey": true }
}

Test that it works

Run this in any terminal. If you see a JSON reply, your key and endpoint are correct.

cURL
curl https://api.unlimitedai.sh/v1/chat/completions -H 'Authorization: Bearer sk-xx' -H 'Content-Type: application/json' -d '{"model":"z-ai/glm-5.3","messages":[{"role":"user","content":"say hi"}]}'

Questions

Where do I get my API key?
Buy any plan — 30 days or a 24-hour pass — and your key is delivered right after payment.
My tool is not on this page
Any tool that accepts a custom OpenAI-compatible base URL works. Enter the endpoint, your key and a model ID — nothing else changes.
Why does a model say not found?
Check the exact model ID on the models page. The 5 exclusive models only work on the All 17 Models plan.
Still stuck?
See the troubleshooting page — it covers 401, 404, 429 and Codex-specific errors in one line each.