OpenCode

OpenCode is an open-source terminal AI coding assistant. With iCreat's OpenAI-compatible API, you can use authorized models inside OpenCode.

Prerequisites

• An iCreat API Key
• The API Key must include the target model, e.g. gpt-5.5 (see available IDs on the Model Plaza)
• OpenCode installed
• Network access to https://api.icreat.ai

Install OpenCode

macOS / Linux

Recommended — official install script:

curl -fsSL https://opencode.ai/install | bash

Or via npm:

npm install -g opencode-ai

On macOS, you can also use Homebrew:

brew install anomalyco/tap/opencode

Windows

OpenCode officially recommends WSL. You can also install with:

npm install -g opencode-ai

Verify installation:

opencode --version

Configure OpenCode

OpenCode supports project-level and global configuration:

• Project-level: opencode.json or opencode.jsonc in the project root
• Global: ~/.config/opencode/opencode.json on macOS / Linux
• Windows: %USERPROFILE%\.config\opencode\opencode.json

Global configuration is recommended so all projects can reuse the same settings. If you use project-level config, do not commit your API Key to the repository.

Config file

Save the following as opencode.json or opencode.jsonc:

{
  "$schema": "https://opencode.ai/config.json",
  "model": "icreat/gpt-5.5",
  "provider": {
    "icreat": {
      "npm": "@ai-sdk/openai-compatible",
      "name": "iCreat AI",
      "options": {
        "baseURL": "https://api.icreat.ai/llm/openai/v1",
        "apiKey": "{env:ICREAT_API_KEY}"
      },
      "models": {
        "gpt-5.5": {
          "name": "GPT-5.5"
        }
      }
    }
  }
}

Field reference:

Note

Use https://api.icreat.ai/llm/openai/v1 for baseURL. Do not use legacy grouped paths, and do not omit the trailing /v1. OpenCode appends paths such as /chat/completions automatically.

Set the API Key

macOS / Linux

export ICREAT_API_KEY="<sk-your-token>"

To persist across terminal sessions, add the line to ~/.zshrc or ~/.bashrc, then run:

source ~/.zshrc

Windows PowerShell

$env:ICREAT_API_KEY = "<sk-your-token>"

To persist on Windows, add a system environment variable named ICREAT_API_KEY, then open a new terminal.

Launch

Start OpenCode from your project directory:

cd your-project-folder
opencode

On first use, run:

/models

Select gpt-5.5 under iCreat AI, or rely on the default model in your config file.

OpenCode also supports /connect for provider credentials, but for iCreat we recommend the ICREAT_API_KEY environment variable — simpler and avoids storing secrets in OpenCode credential files.

Verify the connection

Send this in OpenCode:

Reply with: OpenCode is connected to iCreat AI.

If you receive a normal response, the setup is working.

You can also inspect the loaded config:

opencode debug config

Check Usage Logs in the iCreat console for model, token usage, and billing details.

FAQ

OpenCode cannot find the model

Check:

1. Model IDs in models match iCreat user model codes
2. model uses the correct provider prefix, e.g. icreat/gpt-5.5
3. Your API Key is authorized for that model

401 or Unauthorized

Confirm the API Key is set in the current terminal:

macOS / Linux

test -n "$ICREAT_API_KEY" && echo "ICREAT_API_KEY is set"

Windows PowerShell

if ($env:ICREAT_API_KEY) { "ICREAT_API_KEY is set" }

Also verify the key is not expired, deleted, or wrapped in extra whitespace.

Wrong request URL

baseURL should be:

https://api.icreat.ai/llm/openai/v1

Do not use legacy grouped paths, and do not omit the trailing /v1.

Slow responses

• Confirm your network can reach https://api.icreat.ai
• Try a faster model
• Check service status and usage logs in the iCreat console

Security

• Do not commit API Keys inside opencode.json to Git
• Do not expose API Keys in screenshots, logs, or public chats
• Create a dedicated API Key for OpenCode to track usage and revoke access easily
• If a key is leaked, delete it in the console and create a new one immediately

Related links

• OpenCode installation docs
• OpenCode provider configuration
• OpenCode config reference
• iCreat API Key