Model Gateway Service

The Model Gateway Service serves as the central abstraction layer within the Trimble Agentic AI Platform for interacting with various Large Language Models (LLMs) and other AI models. Its primary purpose is to standardize access, enhance security (content filtering and prompt shielding). It provides a unified interface for agents to other platform components to consume AI model capabilities, regardless of the underlying model provider (OpenAI, Anthropic, Google, Meta, etc.) or inference compute provider (Azure, GCP, etc.). To achieve this, it exposes distinct APIs for model inference (data plane) and service management (control plane).

⚠️ Important: While the Model Gateway Service offers a wide variety of AI Models, you must be aware that larger, smarter models cost more than 10x what smaller, faster models cost to run. Use the minimum size model necessary for your agent to optimize our costs. If you are unsure of what model to use, join our community channel to ask and get advice.

Environment URLs	Inference API   Dev: models.dev.trimble-ai.com   Stage: models.stage.trimble-ai.com   Prod: models.ai.trimble.com Management API   Dev: models-api.dev.trimble-ai.com   Stage: models-api.stage.trimble-ai.com   Prod: models-api.ai.trimble.com
API Docs	developer.stage.trimble-ai.com/#models-inference developer.stage.trimble-ai.com/#models-control-plane
Repository	github.com/Trimble-Agentic/agentic-model-svc
Tech Stack	Python, FastAPI, Pydantic
Database	MongoDB