Llama 4 Scout

Llama 4 Scout is the smaller mixture-of-experts variant in Meta's Llama 4 family, with 17B active parameters across 16 experts and native multimodal support. It is tuned for very long context windows and efficient inference, making it well suited for document-heavy tasks. Use it for long-document RAG, summarization at scale, and multimodal pipelines that demand open weights.

Quick start

Use any OpenAI-compatible client. Just change base_url and your key.

from openai import OpenAI

client = OpenAI(
    api_key="tl-xxxxxxxxxxxxxxxxxxxxxxxx",
    base_url="https://api.thalam.ai/v1",
)

response = client.chat.completions.create(
    model="meta-llama/llama-4-scout-17b-16e-instruct",
    messages=[
        {"role": "user", "content": "Hello from Thalam!"}
    ],
)

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