Model HQ Documentation

Model HQ supports multimodal inference for image analysis, image generation, and text-to-speech. All media endpoints accept prompts and return file paths or streamed text.

Prerequisites

This guide assumes the SDK is installed and importable. If you haven't set that up yet, follow the Hello World prerequisites first.

Image analysis

Non-streaming

Upload an image and ask a question about it:

from modelhq.client import LLMWareClient

client = LLMWareClient(api_endpoint="http://localhost:8088", api_key="")

response = client.vision(
    image_fp="path/to/image.jpg",
    prompt="Describe what is shown in this image"
)
print(response)

Streaming

Stream the analysis token-by-token. There is an initial delay of 10-15 seconds while the image is encoded, then generation is fast:

for token in client.vision_stream(
    image_fp="path/to/image.jpg",
    prompt="Describe what is shown in this image"
):
    print(token, end="")

Image generation

Generate an image from a text prompt. The result is saved to disk and the file path is returned:

save_path = client.generate_image(
    prompt="A watercolour painting of a mountain landscape at sunset",
    file_path="output/landscape.bmp"   # optional — defaults to tmp folder
)
print("image saved to:", save_path)

Text-to-speech

Convert text to speech audio. The result is saved as a .wav file:

save_path = client.generate_speech(
    prompt="Welcome to Model HQ. Let me walk you through the key features.",
    file_path="output/intro.wav"   # optional — defaults to tmp folder
)
print("audio saved to:", save_path)

Web search

Search the web using a configured provider (tavily, serp_api, news_api_org, or wikipedia). Requires a provider API key stored as a credential:

# Store the provider API key
client.add_credential({"tavily_api_key": "tvly-..."})

# Run a search
results = client.web_search(
    query="What is quantum gravity?",
    engine="tavily",
    result_count=5
)
print(results)

Next steps

RAG — combine vision with retrieval for document-heavy workflows
Agents — agents can use vision and search as steps in a pipeline
API Reference — full endpoint specifications

For further assistance or to share feedback, please contact us at support@aibloks.com

Vision & Media — Model HQ Client SDK