AI models keep getting smaller, cheaper and more practical

Small language models are quietly reshaping how businesses actually use AI. Instead of paying for a giant general-purpose model to handle every task, more companies are finding that compact, specialised models do the job at a fraction of the cost. This week the conversation across the industry keeps returning to efficiency, open-source momentum and the practical tools small businesses can use today.

Efficiency is the new arms race

The big frontier model labs still trade headlines about raw capability. But the quieter, more useful trend is running in the opposite direction. Distillation, quantisation and better fine-tuning have all improved, so a model that once needed serious server hardware can now run on a laptop or even a phone. For a small business, that matters. It means an AI assistant can process a proposal, summarise a client brief or draft an email without sending data to a remote data centre and without racking up per-token fees.

This is not a new idea. What has shifted is the quality bar. A couple of years ago, a compact model was noticeably worse than a large one on almost every task. Now the gap has narrowed to the point where, for many narrow, repeated jobs, the small model is good enough and dramatically cheaper. The economics favour it, and the pendulum has swung back toward running models where the data lives.

Open-source momentum keeps building

Open-weight models continue to set the pace for anyone who wants control over cost, privacy and customisation. Because the weights are public, an agency or an in-house team can fine-tune a model on their own documents, their own tone of voice, their own product catalogue. You cannot do that easily with a closed API model. For a boutique agency serving local clients, that control is often the deciding factor.

The practical consequence is that "AI for business" no longer means one vendor, one API key and one way of doing things. It means a toolkit. You can pick a model that fits the job, host it where it makes sense and keep your data where you want it. That flexibility is exactly what small operators in Australia need, because their compliance and privacy obligations do not scale down to match a lower budget.

What this means for a practical business

The takeaway for a tradie, a cafe or a professional services firm is straightforward. You do not need the biggest model to get real value. A focused assistant that understands your business, runs cheaply and keeps your data close is often the better buy than a flashy all-purpose model you rarely use to full depth.

The useful frame is outcome first. Ask what problem you are solving, then pick the smallest tool that reliably solves it. That instinct, choosing the right-sized model for the right job, is becoming the difference between an AI project that pays for itself and one that quietly burns budget.

This article was written by Triweb AI's editorial team based on analysis of today's leading AI news sources.