Claude Cowork
Connect Claude Cowork and Claude Desktop App to ARZASTORE AI API via Third-Party Inference Gateway
Third-Party Inference Gateway
Requirements
Software
- Claude Cowork or Claude Desktop App (latest version)
ARZASTORE AI
- Consumer API Key (sk-xxx)
- Subscribed to Claude model on Marketplace
- Endpoint URL from the Marketplace page
Enable Developer Mode in Claude
- 01Open Claude SettingsGo to Settings (gear icon) → switch to the Account or Advanced tab.
- 02Enable Developer ModeFind the option Enable Developer Mode and enable it. Developer Mode unlocks advanced configuration options.
- 03Restart ClaudeClose and reopen Claude for the changes to take effect.
Go to Settings (gear icon) → switch to the Account or Advanced tab.
Find the option Enable Developer Mode and enable it. Developer Mode unlocks advanced configuration options.
Close and reopen Claude for the changes to take effect.
Safe & Secure
Configure Third-Party Inference Gateway
- 01Open Gateway SettingsIn Developer Mode, go to Settings → Developer → Third-Party Inference Gateway.
- 02Base URLEnter the ARZASTORE AI endpoint URL:https://api.arzastore.com
- 03API KeyPaste your Consumer API Key (from Dashboard → LLM → API Keys):sk-xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
- 04Apply locallyClick the Apply locally button to save the configuration for the current device only.
In Developer Mode, go to Settings → Developer → Third-Party Inference Gateway.
Enter the ARZASTORE AI endpoint URL:
Paste your Consumer API Key (from Dashboard → LLM → API Keys):
Click the Apply locally button to save the configuration for the current device only.
Note on Base URL
Sign In via Gateway
- 01Choose login methodWhen opening Claude, you will see two options: Sign in with Anthropic and Continue with Gateway.Do not select "Sign in with Anthropic" — this will bypass the gateway configuration.
- 02Verify connectionAfter signing in via Gateway, Claude will show a connection indicator. The models listed will be those you have subscribed to on ARZASTORE AI.
When opening Claude, you will see two options: Sign in with Anthropic and Continue with Gateway.
Do not select "Sign in with Anthropic" — this will bypass the gateway configuration.
After signing in via Gateway, Claude will show a connection indicator. The models listed will be those you have subscribed to on ARZASTORE AI.
Select a Model
After connecting, Claude will display a list of models from your ARZASTORE AI subscription. Use the exact model ID or a configured alias:
| Model | Speed | Best for |
|---|---|---|
| claude-haiku-4-5 | Very Fast | Quick replies, simple questions |
| claude-sonnet-4-6 | Medium | Coding, analysis, creative tasks (default) |
| claude-opus-4-7 | Slow | Complex tasks, deep reasoning |
Verify Connection
Send a test request to verify the configuration is working correctly:
curl https://api.arzastore.com/v1/chat/completions -H "Authorization: Bearer sk-your-api-key" -H "Content-Type: application/json" -d '{
"model": "claude-sonnet-4-6",
"messages": [
{"role": "user", "content": "Hello from Claude Cowork!"}
]
}'
Troubleshooting
Error: "Gateway connection failed"
- - Check the Base URL is correct: https://api.arzastore.com (without /v1)
- - Ensure the API Key starts with sk-
- - Check your internet connection and ARZASTORE AI status
Error: "Invalid API Key" or 401
- - Verify the API Key starts with sk- and has not been deleted
- - Check the key on Dashboard → LLM → API Keys
- - Create a new key if needed
Model not showing or 403 error
- - Check that you have subscribed to the Claude model on ARZASTORE AI Marketplace
- - Model ID must match exactly (e.g., claude-sonnet-4-6)
- - Check that your account balance is sufficient
Claude is still using the Anthropic API directly
- - Make sure to choose "Continue with Gateway" when signing in, not "Sign in with Anthropic"
- - Try signing out and signing back in via Gateway
Tips for Effective Use
- -Use the alias sonnet in ARZASTORE AI for automatic failover when a provider fails
- -Use routing strategy fallback to always select the most stable provider
- -Claude Cowork supports Projects and long context — ideal for extended work sessions
- -Track costs via LLM Usage Dashboard to optimize your budget