Providers and integrations
AI Providers
Configure OpenAI, Anthropic Claude, Google Gemini, Ollama, or custom OpenAI-compatible endpoints.
Who this is for
Users connecting Forge to hosted, local, or self-hosted model providers.
Before you start
- Have provider credentials or a local endpoint.
- Know which model you want to use.
- For local/self-hosted models, confirm the server is running.
Step-by-step guide
- 1
Open Settings.
- 2
Choose OpenAI, Anthropic Claude, Google Gemini, Ollama, or Custom OpenAI-compatible.
- 3
Enter credentials, base URL, or local endpoint details.
- 4
Select a known model from dropdowns where available.
- 5
Enter a custom model ID for newer, private, or self-hosted models.
- 6
Test the provider with a small prompt.
- 7
Use stronger structured-output models for Build with AI.
AI Providers
Forge Current step Inputs Output
1
2
Provider settings Configure OpenAI, Anthropic Claude, Google Gemini, Ollama, or custom OpenAI-compatible endpoints.
3
Tips
- Local models vary in speed, context size, and JSON reliability.
- Custom endpoints should follow OpenAI-compatible APIs.
- Build with AI needs reliable JSON output.
Common mistakes / troubleshooting
- Wrong base URL.
- Model ID not available on the selected provider.
- Using a weak structured-output model for kit generation.