Get Started with Kilo Code - Cerebras Inference

Configure the VS Code extension

Open Kilo Code settings

Open the Kilo Code panel in VS Code, select the gear icon, and open the Providers tab.

Add Cerebras

Add Cerebras as a provider and paste your Cerebras API key. You do not need to configure a custom base URL.

Select a model

Choose gemma-4-31b, gpt-oss-120b, or zai-glm-4.7 from the Cerebras model list, then start a coding task.

Configure the CLI

Set your Cerebras API key:

export CEREBRAS_API_KEY="your-api-key"

Add the Cerebras provider and default model to ~/.config/kilo/kilo.json or your project’s kilo.json:

{
  "provider": {
    "cerebras": {
      "env": ["CEREBRAS_API_KEY"]
    }
  },
  "model": "cerebras/gemma-4-31b"
}

Kilo Code refreshes its model catalog automatically. If a newly released model is not visible immediately, restart Kilo Code and check the model selector again.

Available models

Kilo Code supports the models exposed by the Cerebras provider:

Model	Parameters	Best For
gpt-oss-120b	120B	Large coding and agentic tasks
gemma-4-31b	31B	Multimodal coding tasks with text and image inputs
zai-glm-4.7	355B	Complex reasoning and coding tasks

See the model catalog for current availability and the Gemma 4 31B model page for its context, output, and image-input limits.

Troubleshooting

Gemma 4 31B does not appear in the model selector

Confirm that Kilo Code is up to date, restart it to refresh the model catalog, and verify that Cerebras is configured as the provider rather than OpenAI Compatible.

Kilo Code reports an authentication error

Confirm that CEREBRAS_API_KEY is set for the CLI or re-enter your key in the VS Code provider settings. You can create and manage keys in the Cerebras Cloud console.

​Prerequisites

​Configure the VS Code extension

​Configure the CLI

​Available models

​Troubleshooting

Prerequisites

Configure the VS Code extension

Configure the CLI

Available models

Troubleshooting