This week in AI engineering, notable advancements include the Fathom R1 14B model, which successfully passed the IIT JEE Advanced exam. Google’s DeepSearch stack is now open source, allowing for advanced document searching. Additionally, NVIDIA introduced a new reasoning model, the Nemotron Research Qwen 1.5B, designed for a variety of reasoning tasks. Microsoft unveiled a text-to-video generation in Bing, while OpenAI launched new audio tools. The article also highlights emerging tools to enhance development workflows.
Built under India's National AI Mission, Fathom R1 14B is a 14 billion-parameter reasoning model that cracked the IIT JEE Advanced on its first attempt.
Google has open-sourced its entire DeepSearch stack, enabling ultra-fast multimodal document searches with advanced indexing and dual encoders.
NVIDIA's new Nemotron Research Reasoning Qwen 1.5B is a finely-tuned advanced reasoning model designed for complex tasks in various fields.
Microsoft's introduction of Sora-style text-to-video generation in Bing demonstrates significant advancements in AI capabilities for content creation.
Collection
[
|
...
]