Configure Your First Model

Helix is model-agnostic. It manages project context, tool calls, and file changes, but you still need to tell it which AI model to use.

This tutorial connects a model and verifies it with a simple message.

---

Choose a Connection Path

Path A: Use a Helix Account (Fastest)

If you have a Helix account and are signed in, built-in model access is available without setting up API keys.

1. Click the user avatar in the top-right corner, or click Settings in the bottom toolbar
2. Sign in to your Helix account
3. Built-in models appear in the model list automatically

Choose this for the quickest first run

If you just want to complete your first conversation, use the built-in account path. Later, you can add your own provider for cost control, model choice, or private services.

Path B: Bring Your Own API Key

If you already use DeepSeek, OpenAI, Anthropic, or another OpenAI-compatible service, add it manually as a provider.

The example below uses DeepSeek. For other providers, the flow is the same; only the interface type, Base URL, and model ID change.

---

Step-by-Step: DeepSeek Example

Step 1 — Open Settings

You can open settings from either place:

• Click Settings at the far right of the bottom toolbar
• Click the gear icon in the top-right corner

Step 2 — Open LLM Configuration and Add a Provider

In the settings panel, find LLM Configuration or provider management, then click Add Provider.

Step 3 — Fill In Provider Details

For DeepSeek, use values like these:

Field	Value
Name	DeepSeek
Interface type	OpenAI-compatible
Base URL	https://api.deepseek.com
API Key	Your DeepSeek API key

Protect your API key

An API key is a billing credential. Don't share it, don't commit it to a repository, and avoid exposing it in screenshots or recordings. Helix stores it in local configuration, but you should still handle it carefully.

Step 4 — Add a Model

After creating the provider, add a model under it:

Field	Value
Model ID	deepseek-chat
Display name	DeepSeek V3 (customizable)
Context window	65536

Step 5 — Set It as Default

Once the model is added, set it as the default. New sessions will automatically use it.

You can also click the model name in the bottom toolbar at any time to switch between configured models.

---

Verify the Model Works

The simplest check is to send a message:

1. Return to the main screen
2. Click + New Session
3. Type: Hello — introduce yourself in one sentence
4. Press Ctrl+Enter (or Cmd+Enter on Mac) to send

If the AI replies, the model is ready.

If it doesn't reply or shows an error

Check these first:

• API key completeness: no missing characters or extra spaces
• Base URL: providers may require different URL formats
• Model ID: it must match the provider's documentation exactly

---

Common Provider Reference

Provider	Interface type	Base URL	Common Model IDs
OpenAI	OpenAI	https://api.openai.com/v1	gpt-4o, o3
Anthropic	Anthropic	https://api.anthropic.com	claude-sonnet-4-20250514
DeepSeek	OpenAI-compatible	https://api.deepseek.com	deepseek-chat, deepseek-reasoner
SiliconFlow	OpenAI-compatible	https://api.siliconflow.cn/v1	Per platform docs

---

What You Should Have Now

After this tutorial, you should see an available model in the bottom model bar and be able to send a message that receives a reply.

Next Step

With a model ready, start your first conversation and watch Helix read your project, call tools, and produce useful output.