Comet 60B Halves Size of Multimodal AI Model Without Losing Accuracy
News related to:Multiverse Computing · 2 min read
SAN SEBASTIAN, Spain, Oct. 15, 2026 /CourierPR/ -- Computing today announced Comet 60B, a compressed and optimized version of NVIDIA's Nemotron 3.5 Super VL. The original Mixture-of-Experts (MoE) model, launched in June 2026, built on a compressed and optimized version of NVIDIA Nemotron 3 Nano, had 120 billion total parameters and 12 billion active parameters, designed for long, complex tasks and for agentic workloads that combine text, image, and video.
Comet 60B halves the total parameter count to 60 billion, keeps all three modalities, and the same 12 billion active parameters, while improving the original model in some specific areas and benchmarks. For the rest, it stays within 5% of the original model's accuracy across text, image, and video tasks.
Multiverse Computing, which received early access to the model, has cut its size in half while maintaining accuracy within 5% of the original. This allows companies to run Comet 60B on less hardware and at lower cost, making it more accessible and efficient for companies deploying agents.
Comet 60B is designed for teams building and running agentic AI workloads, combining strong coding and agent capabilities with support for text, images, and video. It gives developers a versatile foundation for creating assistants, automating complex tasks, and building applications that work across different types of content.
With 50% fewer parameters, Comet 60B reflects the company's commitment to making advanced AI more efficient and accessible. Available to the community, it enables developers and organizations to explore new multimodal use cases and build AI solutions around their own workflows.
Comet 60B is built on top of a compressed and optimized version of NVIDIA Nemotron 3.5 Super VL. Multiverse Computing produced Comet 60B through its framework CompactifAI. The model is designed to be efficient and accessible, with a focus on reducing the size of the model without sacrificing accuracy.
Key benchmarks show Comet 60B outperforming the original model in some areas, such as the τ²-Bench telecom benchmark, where it scores 83.11 compared to the original's 81.91.
Comet 60B is available today on Hugging Face as BF16, FP8, and NVFP4. Full technical documentation, including methodology, evaluation setup, and per-benchmark results, is available here. To explore Multiverse Computing's open-source model releases, visit huggingface.com/MultiverseComputingCAI.
Comet 60B, Multiverse Computing's new open-source multimodal model, is designed for teams building and running agentic AI workloads. Combining strong coding and agent capabilities with support for text, images, and video, it gives developers a versatile foundation for creating assistants, automating complex tasks, and building applications that work across different types of content.
With 50% fewer parameters, Comet 60B reflects the company's commitment to making advanced AI more efficient and accessible. Available to the community, it enables developers and organizations to explore new multimodal use cases and build AI solutions around their own workflows.
Comet 60B is available today on Hugging Face as BF16, FP8, and NVFP4. Full technical documentation, including methodology, evaluation setup, and per-benchmark results, is available here. To explore Multiverse Computing's open-source model releases, visit huggingface.com/MultiverseComputingCAI.