Long-running agents deserve better inference.
Subconscious

How to use Command Code with Subconscious

Let your AI do the setup. Paste this into Claude Code, Cursor, ChatGPT or any assistant.

Set up Command Code to use Subconscious as its model provider.

Subconscious is an inference API for agents. It serves open models through OpenAI- and Anthropic-compatible endpoints, billed per token.
- OpenAI-compatible base URL: https://api.subconscious.dev/v1
- Anthropic-compatible base URL: https://api.subconscious.dev
- Model: subconscious/glm-5.3-marathon
- API key: read it from the SUBCONSCIOUS_API_KEY environment variable. Never commit the key or paste it into project files. If it is not set, ask me to create one at https://platform.subconscious.dev.

Follow these steps. Run commands and edit files yourself. For any step in a settings screen, tell me exactly what to click and what to enter.

1. Install Command Code
The desktop app for macOS, Windows and Linux includes the agent runtime, so the CLI is optional.
```bash
npm i -g command-code
```

2. Create a Subconscious API key
Sign in at https://platform.subconscious.dev, create an API key, and export it so the tool can read it.
```bash
export SUBCONSCIOUS_API_KEY="your_key"
```

3. Add Subconscious as a provider
apiKey takes a reference to an environment variable, not the key itself. contextWindow defaults to 200K; 5,000,000 matches what the Subconscious CLI uses.
```json
{
  "provider": {
    "subconscious": {
      "name": "Subconscious",
      "baseURL": "https://api.subconscious.dev/v1",
      "apiKey": "$SUBCONSCIOUS_API_KEY",
      "models": {
        "subconscious/glm-5.3-marathon": { "name": "GLM-5.3 Marathon", "contextWindow": 5000000 }
      }
    }
  }
}
```

4. Pick the model
Start Command Code in your project, open /model and choose GLM-5.3 Marathon under the Subconscious heading.
```bash
cmd
```

Full guide: https://www.subconscious.dev/integrations/terminal-and-ide/commandcode.md

When you are done, confirm the tool is using subconscious/glm-5.3-marathon, then summarize what you changed.

To use Command Code with Subconscious, add a subconscious provider to ~/.commandcode/providers.json with base URL https://api.subconscious.dev/v1, your key as $SUBCONSCIOUS_API_KEY and the model subconscious/glm-5.3-marathon. Run cmd, pick the model in /model, and requests go straight to Subconscious.

Get an API keyTeam Token Plans for coding agents

Last verified against the Command Code docs on

Before you start

  • Node.js and npm, or the Command Code desktop app
  • A Subconscious account and API key

Command Code docs · Source on GitHub

Set up Command Code

  1. Step 1: Install Command Code

    The desktop app for macOS, Windows and Linux includes the agent runtime, so the CLI is optional.

    Terminal
    npm i -g command-code

  2. Step 2: Create a Subconscious API key

    Sign in at https://platform.subconscious.dev, create an API key, and export it so the tool can read it.

    Terminal
    export SUBCONSCIOUS_API_KEY="your_key"

  3. Step 3: Add Subconscious as a provider

    apiKey takes a reference to an environment variable, not the key itself. contextWindow defaults to 200K; 5,000,000 matches what the Subconscious CLI uses.

    ~/.commandcode/providers.json
    {
  "provider": {
    "subconscious": {
      "name": "Subconscious",
      "baseURL": "https://api.subconscious.dev/v1",
      "apiKey": "$SUBCONSCIOUS_API_KEY",
      "models": {
        "subconscious/glm-5.3-marathon": { "name": "GLM-5.3 Marathon", "contextWindow": 5000000 }
      }
    }
  }
}

  4. Step 4: Pick the model

    Start Command Code in your project, open /model and choose GLM-5.3 Marathon under the Subconscious heading.

    Terminal
    cmd

Why run Command Code on Subconscious

Long sessions stay fast
2x faster task completion on long tasks than standard open-model inference, especially past 200k tokens of context.
Context that keeps going
The OrangeLine runtime compresses the parts of context that stopped mattering, on the GPU, while the run continues. That gives a 5M+ token context window with no stop-and-compact pause.
Lower cost on long tasks
50% to 80% lower cost on long tasks, because the runtime processes far fewer tokens. Cached input bills at a steep discount, and agentic coding sees cache hit rates above 95%.
Your data stays yours
Prompts, inputs and outputs are not recorded, and nothing you send is used to train a model.

Connection details

For any client that takes a custom OpenAI- or Anthropic-compatible endpoint.

OpenAI-compatible base URL
https://api.subconscious.dev/v1
Anthropic-compatible base URL
https://api.subconscious.dev
API key
From subc login or the Subconscious dashboard
Model
subconscious/glm-5.3-marathon
Multimodal, high-throughput model
subconscious/deepseek-v4.1-flash-marathon

Full reference in the Subconscious API docs.

Questions

Can I add Subconscious without editing providers.json?
Yes. Run /connect, choose (BYOK) Add your own provider, enter subconscious as the id, keep the OpenAI-compatible wire, and enter the Subconscious base URL, your API key and the model ID.
Do requests go through Command Code servers?
No. Command Code's docs say BYOK requests go straight from your machine to the provider's endpoint with your key.
Which Subconscious model should I use with Command Code?
Start with subconscious/glm-5.3-marathon, an open frontier coding model built for long agent runs. subconscious/deepseek-v4.1-flash-marathon is a cheaper, high-throughput alternative that also takes image input. Use the full model ID, including the subconscious/ prefix.
Is Command Code on Subconscious OpenAI- or Anthropic-compatible?
This guide connects Command Code through the OpenAI-compatible Subconscious API. Subconscious serves both formats: OpenAI Chat Completions at https://api.subconscious.dev/v1 and Anthropic Messages at https://api.subconscious.dev.
How is Command Code usage on Subconscious billed?
Per token by default, with no commitment. Heavy users can switch to a Team Token Plan: a large monthly budget with 5x the usage, shared across your team.

More terminal and IDE agents

See all