AI Fact-Checking Tool Boosts Performance with Enhanced Reasoning Settings

PopUpFactCheck.com has announced a significant quality breakthrough following a change in its underlying AI model's operational settings. The platform, which focuses on fact-checking and news verification, had been running on the GPT-OSS-120B model at a medium reasoning effort for several months during its development and testing phases.

Engineers have now switched the model to its high reasoning effort setting. Early results indicate a substantial improvement in the model's capabilities, particularly in complex tasks such as attribution, evidence reconciliation, and judgment, which are crucial for effective fact-checking. The platform's architecture, designed to cache completed fact-checks and utilize cost-effective inference providers, is expected to absorb the increased reasoning costs, leading to a notable enhancement in overall product quality. This upgrade is now live in production.