New AI models drive workplace transformation

The workplace landscape is undergoing a transformation, driven by the unprecedented velocity of artificial intelligence (AI) model development.

- Vishal Mathur

NEW DELHI: While DeepSeek's innovative, frugal approach to developing models has garnered attention, global players continue to make significant strides in making AI more relevant for both consumers and businesses.

This rapid evolution was highlighted when Mistral AI's head of developer relations, Sophia Yang, unveiled the new Saba model on Tuesday, emphasizing global localization. "Mistral Saba is a 24B parameter model trained on meticulously curated datasets from across the Middle East and South Asia," she explained. The model's support for Arabic and multiple Indian languages serves as the foundation for its proficiency in Tamil and Malayalam.

The AI landscape continues to evolve rapidly. OpenAI updated its GPT-40 model, which CEO Sam Altman describes as "pretty good." Meanwhile, Perplexity released its Deep Research tool last week, offering it free of use. "It excels at a range of expert-level tasks—from finance and marketing to product research," according to the AI company. This approach stands in stark contrast to OpenAI's $200 Pro subscription requirement, highlighted by its 03 model.

Voice interactions with AI are becoming increasingly sophisticated. Google Gemini recently added conversational memory retention and 14 new languages for Gemini Live conversations. Its competitor, Microsoft Copilot Voice, expanded its language support to include 40 more languages beyond English, Hindi, and Bengali, while also introducing a 'Think Deeper' feature based on OpenAI's GPT-40 model.