Core42, the AI cloud subsidiary of G42, has introduced OpenAI’s latest open-weight models — GPT-OSS-20B and GPT-OSS-120B — on its AI Cloud platform. Accessible through the Compass API, these models deliver enterprise-grade performance with speeds reaching up to 3,000 tokens per second per user. The deployment runs on diverse GPUs and accelerators, enabling scalable, sovereign AI systems that can be hosted in-country or in the cloud.
For organizations and individuals in the UAE, this launch offers significant advantages:
- Local hosting ensures sensitive data remains in-country and compliant with regulations.
- High-speed performance supports near real-time AI use cases, from customer service to research.
- Hardware flexibility allows optimization across providers and chip types.
- Lower costs make advanced AI accessible to both enterprises and SMEs.
- Custom deployments enable sector-specific solutions in healthcare, finance, education, and public services.
“Core42 AI Cloud, powered by silicon-diverse infrastructure, delivers the flexibility and performance needed for today’s AI workloads,” said Kiril Evtimov, CEO of Core42 and Group CTO of G42. “Through the Compass API, organizations can access the latest open-weight AI models and scale transformation while optimizing performance and cost.”
The open-weight approach empowers businesses with enterprise-scale performance, sovereign-ready scalability, and cost-efficient agentic AI for regulated environments. This milestone reinforces G42’s position as a global AI leader, following the launch of the Stargate UAE facility, its 5GW US-UAE AI campus, and a $1.5 billion investment from Microsoft in 2024.
This initiative marks a turning point for enterprise AI autonomy in the UAE, equipping local organizations with the tools to innovate securely, efficiently, and at scale.
