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

# Integrations

> Connect AI model providers to the Traceport Gateway.

Before routing traffic through Traceport, the Gateway needs authorization to communicate with your downstream LLM providers. The **Integrations** page lets you manage these connections.

Navigate to **Integrations** under API Management in the sidebar.

<Frame>
  <video autoPlay loop muted playsInline src="https://mintcdn.com/traceport/ZXBNThc7kv_3RvjA/images/integrations.mp4?fit=max&auto=format&n=ZXBNThc7kv_3RvjA&q=85&s=de5f0ee2d6d5a3112b0d394df0b280b9" className="w-full rounded-lg" data-path="images/integrations.mp4" />
</Frame>

## Connecting a New Provider

<Steps>
  <Step title="Navigate to Integrations">
    Open the Integrations page and click the **Available** tab to see providers you haven't connected yet.
  </Step>

  <Step title="Select a Provider">
    Click on your intended provider to begin setup.
  </Step>

  <Step title="Add Your API Key">
    Provide your provider API key from their developer console. Traceport securely stores and uses this key to forward requests.
  </Step>

  <Step title="Save">
    Save the connection. Connected providers move to the **Connected** tab.
  </Step>
</Steps>

## Supported Providers

| Provider             | Slug         | Capabilities       | Description                                    |
| -------------------- | ------------ | ------------------ | ---------------------------------------------- |
| **OpenAI**           | `openai`     | Chat, Embed, Batch | GPT-5.4 series, GPT-4o, DALL-E 3               |
| **Anthropic**        | `anthropic`  | Chat               | Claude 4.6 (Opus, Sonnet, Haiku), Claude 3.5   |
| **Google Gemini**    | `google`     | Chat, Embed        | Gemini 3.1 (Pro, Flash), Gemini 1.5            |
| **Google Vertex AI** | `vertex`     | Chat, Embed        | Vertex-hosted foundation models                |
| **Groq**             | `groq`       | Chat, Batch, Files | Ultra-fast LPUs for Llama 4 and Mistral        |
| **Mistral AI**       | `mistral`    | Chat, Embed, Batch | Native flagship models (Large 3, Small 4)      |
| **Together AI**      | `together`   | Chat, Embed        | Large-scale hosting for specialized OSS models |
| **DeepSeek**         | `deepseek`   | Chat               | High-performance reasoning (V3.2, R1)          |
| **Fireworks AI**     | `fireworks`  | Chat, Embed        | Ultra-fast inference for curated open models   |
| **Perplexity**       | `perplexity` | Chat (Citations)   | Search-augmented AI with live citations        |
| **OpenRouter**       | `openrouter` | Chat (Unified)     | Aggregated access to 200+ flagship models      |

## 🤖 Latest Models Catalog (April 2026)

Traceport supports the following frontier models across native and aggregated provider connections:

### OpenAI

* `gpt-5.4-pro`: Next-gen flagship with 2M context window.
* `gpt-5.4-thinking`: Specialized for complex, long-form reasoning.
* `gpt-5.4-mini`: High-speed frontier model for efficient tasks.

### Anthropic

* `claude-4-6-opus`: Multi-modal flagship with 1M context.
* `claude-4-6-sonnet`: The industry standard for balanced speed and intelligence.
* `claude-4-6-haiku`: Ultra-fast multi-modal reasoning.

### Google

* `gemini-3.1-pro`: Advanced multi-modal reasoning with 2M context.
* `gemini-3.1-flash`: Optimized for speed and high-throughput multi-modal tasks.

### Open Source & Specialized Ecosystem

We provide high-performance inference for the leading open-weight models:

* **Llama 4**: Supported via Groq and Together AI (405B Maverick, 70B Scout).
* **DeepSeek**: DeepSeek V3.2 and R1 (Reasoning-focused).
* **Mistral**: Mistral Large 3 and Small 4 native integrations.
* **Perplexity**: Sonar Pro and Sonar Reasoning Pro with live search capability.

## Managing Connected Providers

Connected providers appear in the **Connected** tab. From there you can:

* **Update** API keys if they've been rotated.
* **Disconnect** a provider if it's no longer needed.

<Tip>
  Connect multiple providers to take full advantage of [Configs](/ai-tools/configs) — set up fallback routing so your app automatically switches providers during outages.
</Tip>