> ## Documentation Index
> Fetch the complete documentation index at: https://docs.lucid.foundation/llms.txt
> Use this file to discover all available pages before exploring further.

# Use Hosted Models

> Deploy with OpenAI, Anthropic, and other hosted models via TrustGate

Lucid agents can use any hosted LLM provider without managing infrastructure. TrustGate acts as a unified proxy, routing requests to the right provider while automatically generating cryptographic receipts for every call.

## Supported Providers

TrustGate uses LiteLLM wildcard routing, so any model from a configured provider works immediately -- no config changes needed when providers release new models.

| Provider  | Model Pattern | Example                              |
| --------- | ------------- | ------------------------------------ |
| OpenAI    | `openai/*`    | `openai/gpt-4.1`, `openai/gpt-4o`    |
| Anthropic | `anthropic/*` | `anthropic/claude-sonnet-4-20250514` |
| Google    | `gemini/*`    | `gemini/gemini-2.5-pro`              |
| Mistral   | `mistral/*`   | `mistral/mistral-large-latest`       |
| Groq      | `groq/*`      | `groq/llama-3.1-70b`                 |
| DeepSeek  | `deepseek/*`  | `deepseek/deepseek-chat`             |

Models with `format=api` are always marked as available since they route through TrustGate and require no dedicated compute.

## Launching an Agent with a Hosted Model

Using the CLI (Path B -- no-code):

```bash theme={null}
lucid launch --runtime base \
  --model openai/gpt-4o \
  --prompt "You are a helpful research assistant" \
  --target docker
```

This deploys the pre-built base runtime image which routes inference through TrustGate automatically. You do not need your own API keys -- TrustGate manages provider credentials.

## Using the API Directly

All hosted models are accessible through the OpenAI-compatible inference endpoint:

```bash theme={null}
curl -X POST https://api.lucid.foundation/v1/chat/completions \
  -H "Authorization: Bearer lk_your_api_key" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "openai/gpt-4o",
    "messages": [{"role": "user", "content": "Hello"}]
  }'
```

Unprefixed model names also work for common models (e.g., `gpt-4o` maps to `openai/gpt-4o`).

## Listing Available Models

```bash theme={null}
# All models
curl https://api.lucid.foundation/v1/models \
  -H "Authorization: Bearer lk_your_api_key"

# Only models that can serve inference right now
curl "https://api.lucid.foundation/v1/models?available=true" \
  -H "Authorization: Bearer lk_your_api_key"
```

The `?available=true` filter returns only models that can serve inference now. For hosted models this is always true. For self-hosted models (`format=safetensors` or `gguf`), availability depends on healthy compute nodes with compatible hardware and a recent heartbeat (within 30 seconds).