- MODELS
- ROUTING
- CHAT
- HUMAN
Short answer
How can I switch models without changing API keys?
You use a single gateway endpoint that acts as a proxy for multiple foundation models. Your AI client sends requests to this gateway, which then routes them to the right model. This setup gives you access to the following outcomes.
Where every request lands.
Routing outcomes
Your AI uses the gateway to direct traffic to the right place.
MODELS
Model availability
The AI checks which models are currently active in your gateway. You get a clear list of what's ready to use.
ROUTING
Single endpoint access
Requests go through one point of entry. You avoid the mess of updating different keys for every model change.
CHAT
Direct model interaction
The AI executes chat completions through the gateway. You get responses from the selected model immediately.
HUMAN
Manual key management
The AI cannot create or manage your external provider accounts. You still own the underlying provider relationships.
The workflow
What your AI does when a request arrives.
The AI handles the heavy lifting of model selection and execution.
Inventory models
The AI looks up the available models currently hosted in your TrueFoundry gateway to see what it can use.
truefoundry_list_gateway_modelsSelect target
Based on your instructions, the AI identifies which specific model from the list is the best fit for the task.
truefoundry_list_gateway_modelsExecute chat
The AI sends your prompt to the gateway, which routes the message to the chosen model.
truefoundry_run_gateway_chatReturn response
The AI receives the model's output through the gateway and presents it to you.
truefoundry_run_gateway_chat
Try it
Copy these to start.
Use these prompts to begin interacting with your gateway.
Starting points
These are starting points. You should swap out the model names or specific task details to fit your needs.
TrueFoundry Connector
You're all set. Choose your MCP client and follow the setup instructions.
https://edge.vinkius.com/vk_preview_f00nYwHsDWFv4EirgRZbXCnvKYC62empWE9xeumU/mcpClaude Desktop
Follow the steps below to connect in seconds.
- 1In Claude Desktop, open Settings → Connectors.
- 2Click “Add custom connector” and paste the connector link above as the remote MCP server URL.
- 3Click Add and start a new chat — TrueFoundry capabilities are ready to use.
{
"mcpServers": {
"truefoundry-mcp": {
"url": "https://edge.vinkius.com/vk_preview_f00nYwHsDWFv4EirgRZbXCnvKYC62empWE9xeumU/mcp"
}
}
}List all the models currently available in my TrueFoundry gateway.
Run a chat request using the model named 'llama-3-70b' through the gateway.
Which models can I use right now to process this text?
Send this prompt to the most capable model available in my gateway.
Claude
ChatGPT
Cursor
VS Code
Windsurf
Claude Code
JetBrains
Cline
Start here
Connect TrueFoundry once, then ask.
Follow the link to connect your account. Your credentials stay encrypted, and you can immediately start asking your AI to route requests.
Connect TrueFoundry to your AIFAQ
How the routing works
- 01
Can the AI add new models to my gateway?
No. The AI can only list and use the models you have already configured in TrueFoundry.
- 02
Do I still need to provide my own OpenAI or Anthropic keys?
No. Once you set up the gateway in TrueFoundry, the AI uses the gateway endpoint instead of individual provider keys.
- 03
Can the AI delete a model from my list?
No. The AI has no capability to delete or modify your gateway configuration.
- 04
How does the AI know which models are online?
The AI uses the tool to fetch the current list of active models from your gateway.
- 05
Can the AI switch models mid-conversation?
Yes. You can tell your AI to use a different model from the available list for the next request.
More questions about TrueFoundry? The Connector page answers them. See everything the TrueFoundry Connector can do
Connect TrueFoundry to Claude, Cursor, ChatGPT & more
