Custom API

Use your own AI endpoint or self-hosted model.

Who This Is For

This is for:

• Using self-hosted LLMs (Ollama, LM Studio, etc.)
• Corporate AI deployments with custom endpoints
• Alternative AI providers with OpenAI-compatible APIs

Most users should use OpenAI or Anthropic instead.

Requirements

Your API must be OpenAI-compatible. This means:

• Same endpoint format (/v1/chat/completions)
• Same request/response structure
• Bearer token authentication

Setup

1. Open Tessel
2. Go to Settings
3. Click AI Configuration
4. Click Add API Key
5. Select Custom API
6. Enter:
   • Base URL: Your API endpoint (e.g., http://localhost:11434/v1)
   • API Key: Your authentication token (or leave empty for local)
   • Model: The model name to use
7. Click Save

Examples

Ollama (Local)

Run AI models on your own computer with Ollama.

1. Install Ollama
2. Run a model: ollama run llama3
3. In Tessel settings:
   • Base URL: http://localhost:11434/v1
   • API Key: (leave empty)
   • Model: llama3

LM Studio (Local)

1. Download LM Studio
2. Download a model and start the server
3. In Tessel settings:
   • Base URL: http://localhost:1234/v1
   • API Key: (leave empty)
   • Model: (use the model name from LM Studio)

Azure OpenAI

For corporate Azure deployments:

• Base URL: https://YOUR-RESOURCE.openai.azure.com/openai/deployments/YOUR-DEPLOYMENT
• API Key: Your Azure API key
• Model: Your deployment name

Other Providers

Many providers are OpenAI-compatible:

• Together.ai
• Groq
• Mistral AI
• And more...

Check their docs for the base URL and authentication format.

Troubleshooting

"Connection failed"

• Check the URL is correct
• Make sure the server is running
• Check firewall settings

"Model not found"

• Verify the model name exactly matches
• For Ollama, run ollama list to see available models

Slow responses

Local models depend on your hardware:

• CPU-only: Expect 1-5 seconds per transform
• GPU: Much faster, similar to cloud APIs

Need Help?

Email support@gettessel.com