Is this page helpful?

Hybrid Frontier Model#

The checked-in configs/config_frontier_models.yml profile uses GPT Luna for bounded classification and research roles, GPT Sol for coordination and writing, and NVIDIA NIM for document summaries.

Shipped Configuration#

Role

Model

Intent classification and shallow research

gpt-5.6-luna with role-specific token budgets

Clarification, orchestration, and planning

gpt-5.6-sol

Source routing and research

gpt-5.6-luna

Report writing

gpt-5.6-sol with the writer token budget from the checked-in config

Document summaries

google/gemma-4-31b-it

Use the profile as checked in so role assignments, inference parameters, retry limits, and structured-response behavior stay aligned with the documented configuration. Before deployment, run the complete workflow against the exact provider endpoints and credentials you intend to use.

Important

Other bring-your-own models or modified role assignments are custom profiles outside this documented combination. OpenAI-compatible transport does not guarantee equivalent tool-calling or structured-output behavior. A custom model can require provider-specific prompt, hyperparameter, tool-calling, and structured-output tuning and should be treated as experimental until the complete workflow passes evaluation.

Prerequisites#

  • NVIDIA_API_KEY for the Gemma document-summary model

  • OPENAI_API_KEY for the GPT Sol/Luna roles

  • TAVILY_API_KEY for the default web-search tools

Set these values in deploy/.env; do not store credentials in the YAML file.

Run the Profile#

Start the web API from the repository root:

uv run dotenv -f deploy/.env run nat serve \
  --config_file configs/config_frontier_models.yml

For Docker Compose, set the following value in deploy/.env before starting the stack:

BACKEND_CONFIG=/app/configs/config_frontier_models.yml

Then follow the standard Docker Compose startup procedure.

To create a custom model profile, copy the checked-in configuration, change model references in YAML rather than Python, and evaluate the resulting workflow. Refer to Swapping Models for the support boundary and role-mapping guidance.