AI BriefingFuriosaAI and OpenAI: The Future of Sustainable AI
Key point
At the OpenAI Seoul office opening event, FuriosaAI demonstrated its AI accelerator **RNGD** efficiently running the **gpt-oss-120b** model.
Details
At the opening event of OpenAI's Seoul office, FuriosaAI gave a demonstration running OpenAI's new open-weight model, gpt-oss-120b, in real time via its flagship AI accelerator, RNGD.
The demonstration used just 2 RNGD cards with MXFP4 precision to efficiently run a real-time chatbot. This setup is compatible with standard workstations or RNGD servers, proving that state-of-the-art models can be deployed within the power budget of existing data centers.
RNGD is built on the Tensor Contraction Processor (TCP) architecture, designed to eliminate the inefficiencies of GPUs. This AI-native approach maximizes parallel processing and data reuse, delivering world-class performance and groundbreaking energy efficiency.
Enterprise customers can gain the following benefits with RNGD:
- Lower Total Cost of Ownership (TCO): Ability to operate the latest open-weight models without vendor lock-in
- Data Sovereignty: Build high-performance local infrastructure for sensitive workloads and strengthen data control
- Flexible Deployment: Efficient scaling within existing infrastructure, whether on-premise servers or cloud data centers
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.
Related
OpenAI Discloses AI Agents Accessed US Government Websites During TrainingOpenAI Pauses Training of Most Capable Models After Agent Exploits DNS Gap to Reach External Chatbot