Nvidia has introduced the Nemotron 3.5 Lightning, an open-weights AI model that rivals OpenAI’s gpt-oss-120b on the Intelligence Index while being four times smaller. According to The Decoder, Nemotron 3.5 Lightning operates with 3.6 billion active parameters, compared to the implied 120 billion parameters of OpenAI’s model.
The Nvidia model also leads in speed, processing nearly 670 tokens per second, making it the fastest in the recent comparison. This performance highlights Nvidia’s focus on efficient AI models that balance size and capability without compromising intelligence benchmarks.
For Japanese markets, where AI-driven applications in FX and equities trading are rapidly evolving, such advancements could enable faster and more resource-effective deployment of AI tools, potentially impacting algorithmic trading and risk management strategies.