Press Space to continue
Press Space to continue
Coinbase took a free AI model, trained it on their own fraud data for a few weeks, and it now beats the most expensive model on the market at catching fraud. By a lot. Up to 35 points better on recall. The free model has 9 billion parameters. It runs on a single GPU. Here's why the big model never had a chance. Fraud patterns are proprietary. Coinbase's labeled fraud data has never touched the public internet, so no amount of pretraining compute buys you knowledge of it. Opus 4.5 walked into…
Brian Armstrong@brian_armstrong16h ago
We were able to fine tune an open weight model (Qwen3.5-9B) on proprietary internal data at @coinbase, and it out performed frontier models, at much lower cost and speed, for a specialized task (fraud prevention in this case). I think we'll see more of this pattern inside companies. The frontier models are amazing generalists. But increasingly you'll see specialist models out perform on more narrow tasks, either working on teams, or by themselves. We're making it easy to fine tune models internally for our employees. We even have a skill for it in our internal harness that makes it easy to spin these up and start training them: "/fine-tune-and-serve-llm I want to fine-tune Qwen for payments risk review. help me get started" I imagine we'll have some backend services that could be replaced entirely by this new architecture.
