Welcome back to Tech Brief, where we unpack the week’s biggest stories in artificial intelligence. Today, The Economist argues that “forget DeepSeek: large language models are getting cheaper still.” As recently as 2022, building an LLM “was a feat at the cutting edge of AI engineering.” Fast-forward three years, and the real challenge isn’t just quality—it’s cost. “To really stand out in the crowded marketplace, an AI lab needs not just to build a high-quality model, but to build it cheaply,” the article warns.

Innovation in model pruning, quantization and knowledge distillation has slashed development bills. Open-source contenders like LLaMA and Mistral are driving down inference costs, while custom chips and smarter training recipes churn out new models at a fraction of yesterday’s price. The result? A price war that promises affordable, powerful AI for businesses of all sizes.

That’s your one-minute update on the race to cheaper LLMs. I’m your host, Emma Cross—thanks for listening to Tech Brief. We’ll see you next time.
Link to Article