---
title: "Google and open-weight models | Skills We Assign For | Azendo"
description: "Open-weight models — matching size to task, licensing, deployment, plus the Azendo roles assigned for it."
url: "https://azendo.co/skills/google-and-open-weight-models/"
---

[Skills](https://azendo.co/skills/) AI and machine learning 

# Google and open-weight models.

Open-weight models — Gemma, Llama, Mistral, Qwen and others — are published with downloadable weights. They cover a range of sizes, and the smaller ones run on modest hardware including CPUs and edge devices.

## Where Google and open-weight models fits on a long engagement.

The most useful property of open weights is the size range. Many production tasks — classification, extraction, routing, summarisation of short text — are handled well by a small model that runs cheaply, and using a frontier model for them is paying substantially more for capability the task does not need.

Licensing needs reading rather than assuming. "Open weights" covers a range from genuinely permissive to licences with usage restrictions, scale thresholds or field-of-use limits. A model chosen without checking can become a commercial problem well after it is embedded in a product.

## What an assigned team does with Google and open-weight models.

Model selection should be empirical. A small model evaluated on the actual task frequently matches a much larger one at a fraction of the cost and latency, and that comparison takes days rather than the weeks teams assume.

Running that evaluation properly, rather than defaulting to the largest available, is part of what AI capacity is for under [ai engineering services](https://azendo.co/services/ai-engineering/).

## What we use Google and open-weight models for.

* Small models for bounded tasks Classification and extraction handled at a fraction of frontier-model cost.
* Licences checked before adoption Usage restrictions understood, because they are a commercial problem once embedded.
* On-device and edge inference Models small enough to run where no connection or no data egress is permitted.

## How Google and open-weight models capacity is assigned.

Model evaluation is assigned as explicit scope, with size matched to the task by measurement rather than by defaulting to the largest option.

## Roles we assign Google and open-weight models for

* [AI Engineer AI engineering](https://azendo.co/services/ai-engineering/ai-engineer/)

## Service lines it sits in

* [AI engineering](https://azendo.co/services/ai-engineering/)

Capacity is agreed as a committed monthly capacity across a discipline, not per skill.

## Related in ai and machine learning

* [PyTorch — skill we assign for](https://azendo.co/skills/pytorch/)
* [TensorFlow — skill we assign for](https://azendo.co/skills/tensorflow/)
* [scikit-learn — skill we assign for](https://azendo.co/skills/scikit-learn/)
* [Pandas — skill we assign for](https://azendo.co/skills/pandas/)
* [NumPy — skill we assign for](https://azendo.co/skills/numpy/)
* [embeddings — skill we assign for](https://azendo.co/skills/embeddings/)
* [function calling — skill we assign for](https://azendo.co/skills/function-calling/)
* [vLLM — skill we assign for](https://azendo.co/skills/vllm/)

## Tell us what your roadmap needs Google and open-weight models for.

A service delivery manager replies with the disciplines we would assign, the monthly capacity and what the first month looks like.

[All skills we assign for](https://azendo.co/skills/)