Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.together.ai/llms.txt

Use this file to discover all available pages before exploring further.

Together AI offers day 1 support for the new Llama 4 multilingual vision models that can analyze multiple images and respond to queries about them. Register for a Together AI account to get an API key. New accounts come with free credits to start. Install the Together AI library for your preferred language.

​
How to use Llama 4 Models

from together import Together

client = Together()  # API key via api_key param or TOGETHER_API_KEY env var

# Query image with Llama 4 Maverick model
response = client.chat.completions.create(
    model="meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8",
    messages=[
        {
            "role": "user",
            "content": [
                {"type": "text", "text": "What can you see in this image?"},
                {
                    "type": "image_url",
                    "image_url": {
                        "url": "https://huggingface.co/datasets/patrickvonplaten/random_img/resolve/main/yosemite.png"
                    },
                },
            ],
        }
    ],
)

print(response.choices[0].message.content)

​
Output

The image depicts a serene landscape of Yosemite National Park, featuring a river flowing through a valley surrounded by towering cliffs and lush greenery.

*   **River:**
    *   The river is calm and peaceful, with clear water that reflects the surrounding scenery.
    *   It flows gently from the bottom-left corner to the center-right of the image.
    *   The riverbank is lined with rocks and grasses, adding to the natural beauty of the scene.
*   **Cliffs:**
    *   The cliffs are massive and imposing, rising steeply from the valley floor.
    *   They are composed of light-colored rock, possibly granite, and feature vertical striations.
    *   The cliffs are covered in trees and shrubs, which adds to their rugged charm.
*   **Trees and Vegetation:**
    *   The valley is densely forested, with tall trees growing along the riverbanks and on the cliffsides.
    *   The trees are a mix of evergreen and deciduous species, with some displaying vibrant green foliage.
    *   Grasses and shrubs grow in the foreground, adding texture and color to the scene.
*   **Sky:**
    *   The sky is a brilliant blue, with only a few white clouds scattered across it.
    *   The sun appears to be shining from the right side of the image, casting a warm glow over the scene.

In summary, the image presents a breathtaking view of Yosemite National Park, showcasing the natural beauty of the valley and its surroundings. The calm river, towering cliffs, and lush vegetation all contribute to a sense of serenity and wonder.

​
Llama4 Notebook

If you’d like to see common use-cases in code see our notebook here .

​
Llama 4 Model Details

​
Llama 4 Maverick

  • Model String: meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8
  • Specs:
    • 17B active parameters (400B total)
    • 128-expert MoE architecture
    • 524,288 context length (will be increased to 1M)
    • Support for 12 languages: Arabic, English, French, German, Hindi, Indonesian, Italian, Portuguese, Spanish, Tagalog, Thai, and Vietnamese
    • Multimodal capabilities (text + images)
    • Support Function Calling
  • Best for: Enterprise applications, multilingual support, advanced document intelligence
  • Knowledge Cutoff: August 2024

​
Llama 4 Scout (Deprecated)

Llama 4 Scout (meta-llama/Llama-4-Scout-17B-16E-Instruct) has been deprecated. Use Llama 4 Maverick (meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8) instead.

​
Function Calling

import os
import json
import openai

client = openai.OpenAI(
    base_url="https://api.together.ai/v1",
    api_key=os.environ["TOGETHER_API_KEY"],
)

tools = [
    {
        "type": "function",
        "function": {
            "name": "get_current_weather",
            "description": "Get the current weather in a given location",
            "parameters": {
                "type": "object",
                "properties": {
                    "location": {
                        "type": "string",
                        "description": "The city and state, e.g. San Francisco, CA",
                    },
                    "unit": {
                        "type": "string",
                        "enum": ["celsius", "fahrenheit"],
                    },
                },
            },
        },
    }
]

messages = [
    {
        "role": "system",
        "content": "You are a helpful assistant that can access external functions. The responses from these function calls will be appended to this dialogue. Please provide responses based on the information from these function calls.",
    },
    {
        "role": "user",
        "content": "What is the current temperature of New York, San Francisco and Chicago?",
    },
]

response = client.chat.completions.create(
    model="meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8",
    messages=messages,
    tools=tools,
    tool_choice="auto",
)

print(
    json.dumps(
        response.choices[0].message.model_dump()["tool_calls"],
        indent=2,
    )
)

​
Output

[
  {
    "id": "call_1p75qwks0etzfy1g6noxvsgs",
    "function": {
      "arguments": "{\"location\":\"New York, NY\",\"unit\":\"fahrenheit\"}",
      "name": "get_current_weather"
    },
    "type": "function"
  },
  {
    "id": "call_aqjfgn65d0c280fjd3pbzpc6",
    "function": {
      "arguments": "{\"location\":\"San Francisco, CA\",\"unit\":\"fahrenheit\"}",
      "name": "get_current_weather"
    },
    "type": "function"
  },
  {
    "id": "call_rsg8muko8hymb4brkycu3dm5",
    "function": {
      "arguments": "{\"location\":\"Chicago, IL\",\"unit\":\"fahrenheit\"}",
      "name": "get_current_weather"
    },
    "type": "function"
  }
]

​
Query models with multiple images

Currently this model supports 5 images as input. 
# Multi-modal message with multiple images
response = client.chat.completions.create(
    model="meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8",
    messages=[
        {
            "role": "user",
            "content": [
                {"type": "text", "text": "Compare these two images."},
                {
                    "type": "image_url",
                    "image_url": {
                        "url": "https://huggingface.co/datasets/patrickvonplaten/random_img/resolve/main/yosemite.png"
                    },
                },
                {
                    "type": "image_url",
                    "image_url": {
                        "url": "https://huggingface.co/datasets/patrickvonplaten/random_img/resolve/main/slack.png"
                    },
                },
            ],
        }
    ],
)
print(response.choices[0].message.content)

​
Output

The first image is a collage of multiple identical landscape photos showing a natural scene with rocks, trees, and a stream under a blue sky. The second image is a screenshot of a mobile app interface, specifically the navigation menu of the Canva app, which includes icons for Home, DMs (Direct Messages), Activity, Later, Canvases, and More.

### Comparison:
1. **Content**:
   - The first image focuses on a natural landscape.
   - The second image shows a digital interface from an app.

2. **Purpose**:
   - The first image could be used for showcasing nature, design elements in graphic work, or as a background.
   - The second image represents the functionality and layout of the Canva app's navigation system.

3. **Visual Style**:
   - The first image has vibrant colors and realistic textures typical of outdoor photography.
   - The second image uses flat design icons with a simple color palette suited for user interface design.

4. **Context**:
   - The first image is likely intended for artistic or environmental contexts.
   - The second image is relevant to digital design and app usability discussions.

​
Llama 4 Use-cases

​
Llama 4 Maverick:

  • Instruction following and Long context ICL: Very consistent in following precise instructions with in-context learning across very long contexts
  • Multilingual customer support: Process support tickets with screenshots in 12 languages to quickly diagnose technical issues
  • Multimodal capabilities: Particularly strong at OCR and chart/graph interpretation
  • Agent/tool calling work: Designed for agentic workflows with consistent tool calling capabilities

​
Llama 4 Scout (Deprecated):

Llama 4 Scout has been deprecated. Use Llama 4 Maverick instead for similar capabilities including summarization, function calling, long context RAG, and multilingual support.