Microsoft and SkillOpt: A Revolution for GPT-5.5 with Markdown

Microsoft recently unveiled SkillOpt, an innovative method that promises to transform the training of artificial intelligence models. In partnership with three Chinese universities, the company has managed to improve the performance of GPT-5.5 by 23 points on procedural tasks, thanks to a simplified approach using a Markdown file.

SkillOpt stands out for its ability to optimize instruction documents for AI agents using a simple text file. This lightweight format clarifies instructions and facilitates their understanding by models, while also making the training process more accessible and cost-effective. The optimizations achieved can be transferred between different models, such as Codex and Claude Code, which is crucial in a diverse technological environment.

The potential impact of SkillOpt is significant for the AI sector. By simplifying model optimization, Microsoft could enable broader adoption of these technologies across various fields, from finance to healthcare. Companies could thus access optimized AI solutions without requiring substantial investments in infrastructure.

The tech community has welcomed this announcement with enthusiasm, although some experts emphasize the need to avoid becoming overly reliant on this single method. As AI models become more accessible, discussions about the ethical and regulatory implications of their use may emerge, prompting regulators to establish clear guidelines.