Supported Models

SambaNova Documentation SambaNova Cloud
1

Access DeepSeek R1, Llama 3.3, 3.2, and 3.1 family of models, along with the Qwen2.5 family of models at full precision via the SambaNova Cloud API!

All models are available to all tiers.

Model details:

DeepSeek-R1 family

  1. DeepSeek Distill R1 Llama 70B
    • Model ID: DeepSeek-R1-Distill-Llama-70B
    • Developer: DeepSeek AI
    • Context length: 16k tokens
    • Model card: View on Hugging Face

Tulu 3 family

  1. Llama 3.1 Tulu 3 405B
    • Model ID: Llama-3.1-Tulu-3-405B
    • Developer: Allen Institute for AI
    • Context length: 16k tokens
    • Model card: View on Hugging Face

Qwen 2.5 family

  1. Qwen2.5 Coder 32B
    • Model ID: Qwen2.5-Coder-32B-Instruct
    • Developer: Qwen
    • Context length: 16k tokens
    • Model card: View on Hugging Face
  2. QwQ 32B Preview
    • Model ID: QwQ-32B-Preview
    • Developer: Qwen
    • Context length: 16k tokens
    • Model card: View on Hugging Face
  3. Qwen2.5 72B
    • Model ID: Qwen2.5-72B-Instruct
    • Developer: Qwen
    • Context length: 16k tokens
    • Model card: View on Hugging Face
  4. Qwen2 Audio [1]
    • Model ID: Qwen2-Audio-7B-Instruct
    • Developer: Qwen
    • Context length: 16k tokens
    • Model card: View on Hugging Face

Llama 3.3 family

  1. Llama 3.3 70B
    • Model ID: Meta-Llama-3.3-70B-Instruct
    • Developer: Meta
    • Context length: 16k tokens (Support up to 128k in Beta. Please contract our support to get early access)
    • Model card: View on Hugging Face

Llama 3.2 family

  1. Llama 3.2 1B
    • Model ID: Meta-Llama-3.2-1B-Instruct
    • Developer: Meta
    • Context length: 16k tokens
    • Model card: View on Hugging Face
  2. Llama 3.2 3B
    • Model ID: Meta-Llama-3.2-3B-Instruct
    • Developer: Meta
    • Context length: 8k tokens
    • Model card: View on Hugging Face
  3. Llama 3.2 11B Vision
    • Model ID: Llama-3.2-11B-Vision-Instruct
    • Developer: Meta
    • Context length: 4k tokens
    • Model card: View on Hugging Face
  4. Llama 3.2 90B Vision
    • Model ID: Llama-3.2-90B-Vision-Instruct
    • Developer: Meta
    • Context length: 4k tokens
    • Model card: View on Hugging Face

Llama 3.1 family

  1. Llama 3.1 8B

    • Model ID: Meta-Llama-3.1-8B-Instruct
    • Developer: Meta
    • Context length: 16k tokens
    • Model card: View on GitHub

  2. Llama 3.1 70B

    • Model ID: Meta-Llama-3.1-70B-Instruct
    • Developer: Meta
    • Context length: 128k tokens
    • Model card: View on GitHub

  3. Llama 3.1 405B

    • Model ID: Meta-Llama-3.1-405B-Instruct
    • Developer: Meta
    • Context length: 16k tokens
    • Model card: View on GitHub

  4. Llama Guard 3 8B

  1. Explanation of Qwen2 Audio The Qwen2 Audio refers to a specific artificial intelligence model developed by Qwen, designed to process and generate audio-related content. With a model ID of Qwen2-Audio-7B-Instruct, it is part of the Qwen 2.5 family of models and has a context length of 16k tokens. This model is available via the SambaNova Cloud API, suggesting its application in tasks that require audio understanding or generation, such as voice assistants, audio transcription, or music synthesis. The inclusion of “Audio” in its name distinguishes it from other models in the Qwen2.5 family that might be more focused on text-based tasks. (Explanation by AI) ↩︎

12 Likes