Multiverse Computing, the European leader in compressed AI models, today announced the release of two breakthrough models: BlackStar 12B, a compressed and optimized OpenAI GPT-OSS-20B, and Hypernova 60B, a compressed and optimized GPT-OSS-120B.
These new models redefine what is possible with compressed AI, delivering state-of-the-art performance at a fraction of the cost and compute footprint of traditional large-scale models.
A New #1 in Quality-to-Cost Performance
The BlackStar 12B AI model by Multiverse Computing now stands as the highest-performing and most cost-efficient AI model in its class, offering the best quality / cost / performance ratio available on the market today.
Designed for enterprise deployments where efficiency and accuracy are equally critical, it enables organizations to run advanced AI workloads with dramatically lower infrastructure requirements.
Outperforming Mistral AI’s Ministral 14B Across Every Key Metric
In benchmark comparisons against Ministral 14B, a similar model from European provider Mistral AI, BlackStar 12B delivered superior results across the board:
BlackStar 12B by Multiverse Computing vs. Ministral 14B by Mistral AI
- 2× better reasoning performance
- +33% higher intelligence index
- 54% lower memory usage
- +47% higher throughput
- +32% faster latency
BlackStar 12B (CAI-GPT-OSS-20B Slim)
BlackStar 12B (CAI-GPT-OSS-20B Slim)
A New Standard for Large-Scale AI Efficiency with Hypernova 60B
Multiverse Computing’s BlackStar 12B model sets an entirely new benchmark when compared to Mistral 3 Large, delivering superior intelligence, dramatically faster performance, and a fraction of the memory footprint. It is the clear leader in its class for organizations requiring top-tier performance without top-tier compute costs.
Hypernova 60B vs. Mistral 3 Large
- 1.9× better reasoning performance
- +18% higher intelligence index
- 68% lower memory usage
- 3× higher throughput
- +67% faster latency
This performance profile makes theHypernova 60B ideal for enterprises, research labs, and public institutions seeking to deploy cutting-edge AI systems with unmatched efficiency.
Hypernova 60B (CAI-GPT-OSS-120B Slim)
These results confirm Multiverse Computing’s leadership in the next generation of efficient AI models that are smaller, faster, cheaper to operate, yet more capable than established alternatives.
By redefining efficiency without compromising intelligence, Multiverse Computing is setting the standard for Europe’s AI future.