Phala Cloud Documentation — Private AI on TEE

For tools that support an OpenAI-compatible provider, use:

Setting	Value
Provider type	OpenAI compatible
Base URL	`https://inference.phala.com/v1`
API key	Your Phala API key
Model	`z-ai/glm-5.3`, or another chat model from `/v1/models`

Create a key under Private AI → API Keys in the Phala dashboard. For the shell examples below, set PHALA_AI_API_KEY in your local environment. Every model on this endpoint runs in a TEE. These clients do not verify attestation locally; use Private AI Proxy when you need a verified channel on your machine.

Continue

Add a model to your local ~/.continue/config.yaml using Continue’s OpenAI-compatible configuration. Keep any other models already in the file:

name: Phala
version: 0.0.1
schema: v1
models:
  - name: Phala GLM 5.3
    provider: openai
    model: z-ai/glm-5.3
    apiBase: https://inference.phala.com/v1
    apiKey: YOUR_PHALA_API_KEY

Zed

Add this provider under language_models in your settings file using Zed’s OpenAI-compatible endpoint configuration. Enter your Phala API key through Settings → AI → LLM Providers:

{
  "language_models": {
    "openai_compatible": {
      "phala": {
        "api_url": "https://inference.phala.com/v1",
        "available_models": [
          {
            "name": "z-ai/glm-5.3",
            "max_tokens": 1048576
          }
        ]
      }
    }
  }
}

GitHub Copilot (VS Code BYOK)

Run Chat: Manage Language Models, choose Add Models → Custom Endpoint, name the group Phala, enter your Phala API key, and select Chat Completions. In the generated chatLanguageModels.json, keep the API-key input reference created by VS Code and configure the model as below. See the Custom Endpoint documentation. This example uses the model’s context window from the catalog and reserves 65,536 tokens for output, within the model’s 1,048,576-token output maximum:

[
  {
    "name": "Phala",
    "vendor": "customendpoint",
    "apiKey": "${input:phalaApiKey}",
    "models": [
      {
        "id": "z-ai/glm-5.3",
        "name": "Phala GLM 5.3",
        "url": "https://inference.phala.com/v1/chat/completions",
        "toolCalling": true,
        "contextWindow": 1048576,
        "maxOutputTokens": 65536
      }
    ]
  }
]

Select the model in chat. For Copilot Business or Enterprise, your administrator can disable this with the Bring Your Own Language Model Key in VS Code policy.

Aider

Use Aider’s OpenAI-compatible endpoint support. The openai/ prefix tells Aider which API format to use:

export OPENAI_API_BASE="https://inference.phala.com/v1"
export OPENAI_API_KEY="$PHALA_AI_API_KEY"
aider --model openai/z-ai/glm-5.3

Goose

Save the following as phala.json in Goose’s custom_providers directory: ~/.config/goose/custom_providers/ on macOS/Linux, or %APPDATA%\Block\goose\config\custom_providers\ on Windows. If you set GOOSE_PATH_ROOT, use <GOOSE_PATH_ROOT>/config/custom_providers/ instead. Create the directory if needed:

{
  "name": "phala",
  "engine": "openai",
  "display_name": "Phala",
  "description": "OpenAI-compatible confidential inference from Phala.",
  "api_key_env": "PHALA_AI_API_KEY",
  "base_url": "https://inference.phala.com/v1",
  "dynamic_models": true,
  "models": [],
  "supports_streaming": true,
  "model_doc_link": "https://docs.phala.com/phala-cloud/confidential-ai/confidential-model/confidential-ai-api",
  "setup_steps": [
    "Sign in to the Phala dashboard",
    "Open Private AI, then API Keys, and create an API key",
    "Copy the key and paste it above"
  ]
}

With PHALA_AI_API_KEY set, run goose configure, choose Configure Providers → Phala, and select z-ai/glm-5.3. Dynamic discovery also lists the non-chat jaredpalmer/kev-4b decision model; choose a chat model.

AnythingLLM

In Settings → AI Providers → LLM, select Generic OpenAI and enter these provider settings, then select Save changes. Use the model’s context_length from the catalog when choosing another model:

Base URL: https://inference.phala.com/v1
API Key: Your Phala API key
Selected Model: z-ai/glm-5.3
Model context window: 1048576

Cherry Studio

Open Settings → Model Services → + Add Provider and add a custom provider with the values below. Enter the base URL in the OpenAI-compatible endpoint/API address row. Select Add, then Get model list and add z-ai/glm-5.3:

Provider Name: Phala
API Key: Your Phala API key
Base URL: https://inference.phala.com/v1

LobeHub

Open App Settings → AI Providers and create a custom AI provider with the values below. Enable the provider and fetch or add z-ai/glm-5.3 in its model list, then select it in your Agent settings:

Provider ID: phala
Provider Name: Phala
Request Format: openai
Proxy URL: https://inference.phala.com/v1
API Key: Your Phala API key

Jan

In Jan Desktop, open Settings → Model Providers → Add Provider, choose OpenAI-compatible, and set Provider name to Phala, Base URL to https://inference.phala.com/v1, and API key to your Phala key. Select Create, then select z-ai/glm-5.3 from the model list. For tool use, edit the model and enable Tools under Capabilities. For Jan Agent CLI, save a provider and start Jan with it:

jan config set --provider phala \
  --api-key "$PHALA_AI_API_KEY" \
  --base-url https://inference.phala.com/v1 \
  --model z-ai/glm-5.3
jan --provider phala --model z-ai/glm-5.3

​Continue

​Zed

​GitHub Copilot (VS Code BYOK)

​Aider

​Goose

​AnythingLLM

​Cherry Studio

​LobeHub

​Jan