OpenAI

OpenAI’s GPT models are among the most widely used language models available. Selu supports all GPT models accessible through the OpenAI API, including the latest GPT-4o variants.

Getting an API key

1. Sign up or log in at platform.openai.com.
2. Go to API Keys in your account settings.
3. Click Create new secret key, name it “Selu”, and copy the key immediately.

Caution

OpenAI only shows the key once. If you lose it, you’ll need to create a new one.

Connecting to Selu

1. Open the Selu dashboard and go to Settings → LLM Providers → OpenAI.
2. Paste your API key. It will be encrypted at rest.
3. Click Test Connection to confirm access.
4. Save your settings.

Choosing a model

GPT-4o

OpenAI’s flagship model. Fast, capable, and supports text and images. Recommended for most use cases.

GPT-4o mini

A smaller, faster, and cheaper version of GPT-4o. Great for everyday tasks where you want quick responses at lower cost.

o1

Optimized for reasoning and complex problem-solving. Best for math, coding, and multi-step analysis.

Select your model under Settings → LLM Providers → OpenAI → Model.

Configuration options

Setting	Description	Default
Max tokens	Maximum length of the agent’s response	4096
Temperature	Controls creativity vs. predictability (0.0–2.0)	0.7
Organization ID	Optional — if your key belongs to multiple orgs	—

Adjust these under Settings → LLM Providers → OpenAI → Advanced.

Using an OpenAI-compatible API

If you’re running a model server that exposes an OpenAI-compatible API (like LM Studio or vLLM), you can point Selu to it by changing the Base URL in the advanced settings:

http://your-server:8080/v1

This works with any server that implements the OpenAI chat completions endpoint.

Troubleshooting

• “Insufficient quota” — Check your billing settings at platform.openai.com/billing. You may need to add a payment method or increase your spending limit.
• Model not available — Some models require API access approval. Check your account’s model access page.
• Timeout errors — Long responses with high max token limits can sometimes time out. Try reducing the max tokens setting.