integration

Use Mistbeam with the tools you already use. Mistbeam exposes an OpenAI-compatible API, so you can point supported coding agents and applications at one endpoint and change providers without changing your workflow.

before you start

1. Create an account.
2. Add credit from your dashboard.
3. Create an API key in the dashboard. Copy it when it is shown because it cannot be viewed again.

Use these values in your client:

API base URL: https://api.mistbeam.com/v1
Model: qwen-3.8-27b
API key: mb_your_key_here

Keep your API key private. Do not commit it to a repository or expose it in browser-side code.

Codex

Add a Mistbeam provider to ~/.codex/config.toml:

model = "qwen-3.8-27b"
model_provider = "mistbeam"

[model_providers.mistbeam]
name = "Mistbeam"
base_url = "https://api.mistbeam.com/v1"
env_key = "MISTBEAM_API_KEY"
wire_api = "responses"

Then set your key before starting Codex:

export MISTBEAM_API_KEY=mb_your_key_here
codex

Pi

Add Mistbeam as a custom provider in ~/.pi/agent/models.json:

{
  "providers": {
    "mistbeam": {
      "baseUrl": "https://api.mistbeam.com/v1",
      "api": "openai-responses",
      "apiKey": "mb_your_key_here",
      "models": [
        { "id": "qwen-3.8-27b" }
      ]
    }
  }
}

Select the model from Pi after restarting it. Store this file securely because it contains your API key.

OpenClaw

Add a custom provider to your OpenClaw configuration:

{
  models: {
    providers: {
      mistbeam: {
        baseUrl: "https://api.mistbeam.com/v1",
        apiKey: "mb_your_key_here",
        api: "openai-responses",
        models: [
          { id: "qwen-3.8-27b", name: "Qwen 3.8 27B" }
        ]
      }
    }
  }
}

Set mistbeam/qwen-3.8-27b as the model in OpenClaw. The exact location of this configuration depends on how your OpenClaw agent is set up.

Hermes Agent

Run the interactive model setup:

hermes model

Choose Custom endpoint, then enter:

API base URL: https://api.mistbeam.com/v1
API key: mb_your_key_here
Model: qwen-3.8-27b

Hermes can use any endpoint that implements OpenAI Chat Completions. Mistbeam also provides the OpenAI Responses endpoint for clients that support it.

Claude Code

Claude Code uses the Anthropic Messages API. Mistbeam’s current public gateway exposes OpenAI-compatible Chat Completions and Responses endpoints, so Claude Code cannot connect directly without an Anthropic-compatible adapter.

If you are using an adapter, configure the adapter with the Mistbeam base URL, API key, and model above. Direct Claude Code setup will be documented when an Anthropic-compatible Mistbeam endpoint is available.

Test the connection

You can verify your key and model with curl:

curl https://api.mistbeam.com/v1/chat/completions 
  -H "Authorization: Bearer $MISTBEAM_API_KEY" 
  -H "Content-Type: application/json" 
  -d '{
    "model": "qwen-3.8-27b",
    "messages": [
      {"role": "user", "content": "Say hello in one sentence."}
    ]
  }'

Switch back

Mistbeam does not lock your configuration. Change the provider’s base URL, API key, and model to return to another provider. Your application or agent workflow can stay the same.