Configure Model Providers

It is recommended to use CC Switch to complete the configuration — an open-source provider management tool with a visual interface, so there is no need to manually edit JSON or environment variables.

1. Install CC Switch

macOS

Or go to Releases to download the .dmg file for manual installation.

Windows

Go to Releases to download the .msi installer.

Linux

2. Add the api.tokenhotAI Provider

Step 1: Add a New Provider

Open CC Switch, go to the provider management page, and click the + button in the top-right corner.

Step 2: Fill in the Configuration

Fill in the fields as shown below, then click + Add to complete.

Config Item	Value	Description
❶ Provider Name	`api.tokenhot.ai-claude`	Custom
❷ Official Website		Provider official website
❸ API Key	Your api.tokenhotAI API Key	Get it from api.tokenhot.ai
❹ Request URL		Do not add a trailing slash
❺ API Format / Auth Field	`Anthropic Messages (Native)` / `ANTHROPIC_AUTH_TOKEN (Default)`	Keep the default
❻ Model Configuration	Leave blank	Leave blank to use the default Claude model
❼ Options	Check "Hide AI Signature" and "Teammates Mode"	Check as needed
❽ Write to General Configuration	✅ Checked	Write to the global configuration so it takes effect for all projects

Step 3: Enable the Provider

After adding it, go back to the list, select api.tokenhot.ai-claude, and click the Use button. When you see the "Switched successfully" message, the setup is complete.

3. Verify the Configuration

Enter the project directory and start Claude Code:

On first launch, press Esc to skip login. After startup, check the top information bar and confirm that API Usage Billing (❶) is displayed, which means billing is going through api.tokenhot.ai:

If all of the above information is correct, the configuration is successful.

4. Use Other Models

Through api.tokenhotAI, you can use non-Claude models in Claude Code (such as Qwen, GLM, DeepSeek, etc.). You only need to filter compatible models in the model marketplace and then configure model mapping.

Step 1: Filter Compatible Models

Go to the tokenho Model Marketplace, and under API Protocol on the left, select Anthropic to filter models that support the Anthropic protocol.

Step 2: Copy the Model Name

Click the target model card. There is a copy button next to the model name, which lets you copy the model ID with one click (for example, claude-opus-4.7).

Step 3: Configure Model Mapping

Go back to CC Switch and find the Model Mapping section in the provider configuration. Fill the copied model name into the corresponding model slot. You can browse available models in the tokenhot Model Marketplace:

Mapping Slot	Description
Primary Model	The model Claude Code uses by default
Thinking Model	The model used for deep reasoning
Default Haiku Model	The model for lightweight tasks
Default Sonnet Model	The model for medium-complexity tasks
Default Opus Model	The model for high-complexity tasks

Troubleshooting

If you encounter issues, run /status in the Claude Code terminal to check the configuration:

/status

Confirm the following key fields:

❶ Anthropic base URL is https://api.api.tokenhot.ai/anthropic

❷ Model shows the model currently in use

Q: It says "Authentication error"

Confirm that the Anthropic base URL is correct and check whether the API Key is valid.

Q: Connection timeout

Confirm that the request URL is set to https://api.api.tokenhot.ai without a trailing slash.

Q: Streaming response is lagging

api.tokenhotAI already includes global acceleration. This is usually a local network issue, so check your local connection and try again.

Configure Model Providers

1. Install CC Switch#

macOS#

Windows#

Linux#

2. Add the api.tokenhotAI Provider#

Step 1: Add a New Provider#

Step 2: Fill in the Configuration#

Step 3: Enable the Provider#

3. Verify the Configuration#

4. Use Other Models#

Step 1: Filter Compatible Models#

Step 2: Copy the Model Name#

Step 3: Configure Model Mapping#

Troubleshooting#