​
Official Models

We offer a range of models, however, our official models are designed to be the most efficient and accessible choices for various AI services. Below is a list of the eight Official Models currently available. These models are known for their low cost and optimized latency because we host them on high-performance, expensive servers. As a result, we highly recommend opting for these models for the best combination of performance and price.
  1. Shuttle 3 (64k)
    • Type: Text Generation
    • Description: Our flagship text generation model, with a 64k token context window, ideal for large, complex tasks.
  2. Shuttle 3 Mini (64k)
    • Type: Text Generation
    • Description: A compact version of Shuttle 3 with the same 64k context window but more affordable, designed for smaller-scale tasks.
  3. Shuttle RemoveBG
    • Type: Background Removal
    • Description: Fast and precise image background removal at an affordable cost.
  4. s1 🌶
    • Type: Text Generation
    • Description: Our high-performing s1 model, designed for complex and nuanced text generation tasks. Able to execute python, browse the web and more while generating a response.
  5. s1 Mini 🌶
    • Type: Text Generation
    • Description: A scaled-down version of the s1 model, offering the same quality at a more affordable rate, excellent for simpler tasks.
  6. Shuttle 3 Diffusion
    • Type: Image Generation
    • Description: Our go-to, high-quality image generation model based on diffusion techniques—ideal for all image-generation needs.
  7. Shuttle Jaguar
    • Type: Image Generation
    • Description: Our high-quality image generation model perfect for photo realistic images.
  8. Shuttle STT
    • Type: Audio Transcription
    • Description: Highly accurate speech-to-text transcription at a low cost and ideal for converting audio into text with minimal latency.
By using these models, you not only benefit from low costs and high-speed processing but also from the stability we provide through self-hosting on cutting-edge servers. For cost optimization and performance improvements, we encourage users to prioritize these official models wherever possible.

​
Get Models via API

To get a list of available models, you can make a GET request to: https://api.shuttleai.com/v1/models The response will return models in the following format: 
{
  "object": "list",
  "data": [
    {
      "id": "shuttleai/shuttle-3",
      "object": "model",
      "created": 1727049600,
      "owned_by": "shuttleai",
      "pricing": {
        "type": "tokens",
        "in_cost_per_million": 0.5,
        "out_cost_per_million": 0.7
      }
    },
    {
      "id": "shuttle-2-turbo",
      "object": "proxy",
      "proxy_to": "shuttleai/shuttle-3"
    },
    // ... and much more
  ]
}
The proxy_to field in models means that requests sent to shuttle-2-turbo will automatically be forwarded to shuttleai/shuttle-3.
For extra details about each model, you can make a GET request to: https://api.shuttleai.com/v1/models/verbose The response will return detailed information in the following format: 
{
  "object": "list",
  "data": [
    {
      "id": "shuttleai/shuttle-3",
      "object": "model",
      "created": 1727049600,
      "owned_by": "shuttleai",
      "name": "Shuttle 3 (64k)",
      "description": "ShuttleAI's smartest, most powerful model. Shuttle-3 offers highly competitive performance, outperforms Llama-3-405B and Gemini-1.5-Pro.",
      "type": "chat.completions",
      "modality": "text,image->text",
      "permission": {
        "allow_max_context_tokens": 65536,
        "allow_max_output_tokens": 8192,
        "allow_tool_calling": true,
        "allow_streamed_tool_calling": true,
        "allow_parallel_tool_calling": true,
        "allow_internet_param": true,
        "allow_tone_param": false,
        "allow_citations_param": false,
        "training_data_cutoff": "12-2023"
      },
      "active": true,
      "pricing": {
        "type": "tokens",
        "in_cost_per_million": 0.5,
        "out_cost_per_million": 0.7
      }
    }
    // ... and much more
  ]
}