Japan’s Preferred Networks (PFN) has launched "SLM Customize," a web service allowing users to create personalized Small Language Models (SLMs) without any coding. Historically, custom AI required expensive hardware and deep technical expertise. This platform automates the process; users simply input character traits, and the system generates training data and fine-tunes the model. By removing these barriers, PFN makes advanced AI development accessible to a wider audience, enabling anyone to design unique digital personalities.

The training utilizes PFN's proprietary MN-Core AI chips, completing complex model optimization in just 45 minutes. Users can monitor progress through real-time graphs instead of code, making the experience intuitive. The system also applies quantization, a technique that reduces model size and memory usage while maintaining performance. This ensures the AI is efficient enough to run on standard hardware, effectively bridging the gap between high-end server performance and personal computing convenience.

A standout feature is the ability to download finished models for local storage and offline use. This provides users with permanent ownership and eliminates the need for persistent cloud connectivity or expensive infrastructure. By transforming AI development into a user-friendly experience, PFN is fostering a culture where individuals can possess AI reflecting their own values. This shift empowers users to "tame" and own digital assets that are truly unique to their personal or creative needs.

Japan’s Preferred Networks (PFN) has launched "SLM Customize," a web service allowing users to create personalized Small Language Models (SLMs) without any coding. Historically, custom AI required expensive hardware and deep technical expertise. This platform automates the process; users simply input character traits, and the system generates training data and fine-tunes the model. By removing these barriers, PFN makes advanced AI development accessible to a wider audience, enabling anyone to design unique digital personalities.

The training utilizes PFN's proprietary MN-Core AI chips, completing complex model optimization in just 45 minutes. Users can monitor progress through real-time graphs instead of code, making the experience intuitive. The system also applies quantization, a technique that reduces model size and memory usage while maintaining performance. This ensures the AI is efficient enough to run on standard hardware, effectively bridging the gap between high-end server performance and personal computing convenience.

A standout feature is the ability to download finished models for local storage and offline use. This provides users with permanent ownership and eliminates the need for persistent cloud connectivity or expensive infrastructure. By transforming AI development into a user-friendly experience, PFN is fostering a culture where individuals can possess AI reflecting their own values. This shift empowers users to "tame" and own digital assets that are truly unique to their personal or creative needs.