Model HQ Documentation

Model HQ Apps combine agents, models, and configurations into a packaged experience. The chat API lets you interact with an app through a session, sending messages and receiving streamed responses.

Prerequisites

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

Quick example

from modelhq.client import LLMWareClient

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

# Start a session for an installed app
session = client.start_app_session("contract_review")
session_id = session.get("session_id", "")
print("session started:", session_id)

# Send a message and stream the response
for step in client.chat(session_id=session_id, prompt="What is the base salary?"):
    print(step)

# Stop the session when done
client.stop_app_session(session_id)

Starting a session

start_app_session creates a new session ID and instantiates the app state on the server. The app must already be installed on the server.

session = client.start_app_session("app_name")
session_id = session.get("session_id", "")

Optional parameters let you preload models or source documents:

session = client.start_app_session(
    "app_name",
    preload_models=True,
    model_name="llama-3.2-3b-instruct-ov",
    preload_source=True
)

Sending messages

chat sends a user message and returns the response. By default it streams tokens line-by-line:

for step in client.chat(session_id=session_id, prompt="Summarise the document"):
    print(step)

To get a single non-streaming response, set stream=False:

response = client.chat(session_id=session_id, prompt="Summarise the document", stream=False)
print(response)

Running an agent inside a session

Some apps expose agents that can be run within the session context. run_app_agent executes the agent and streams step-by-step output:

for step in client.run_app_agent(
    session_id=session_id,
    agent_name="contract_analyzer",
    agent_inputs={"User-Document": "contract.pdf"}
):
    print(step)

Managing sessions

Check session state

state = client.get_app_state(session_id)
print(state)

Export a transcript

The transcript can be saved as docx, pptx, xlsx, csv, json, or txt:

result = client.get_app_transcript(session_id, output_format="docx")
print("transcript saved to:", result["response"]["zip_fp"])

Stop a session

client.stop_app_session(session_id)

Listing and discovering apps

# List all available apps
apps = client.get_all_apps()
for app in apps.get("app_list", []):
    print(app)

# Get details for a specific app
details = client.lookup_app("contract_review")
print(details)

Next steps

Agents — discover and run standalone agent processes
RAG — build knowledge bases and ask questions over documents
API Reference — full endpoint specifications

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

Chat — Model HQ Client SDK