ModelNova Fusion Studio Revolutionizes Arm Edge AI Development
March 14, 2026, 10:14 am
ModelNova Fusion Studio transforms Edge AI development for Arm-based silicon. embedUR systems announced significant Arm ecosystem expansion. The desktop application streamlines complex workflows. Developers now access native ExecuTorch, Arm Ethos NPU, SDS Framework, and Keil MDK integrations. This powerful platform offers an end-to-end MLOps workflow. It includes over 160 pre-trained models and AI assistance. Fusion Studio cuts cloud costs. It accelerates deployment for diverse Arm targets. This new release simplifies Edge AI projects.
Edge AI development faces ongoing challenges. Fragmented workflows hinder innovation. Developers assemble many tools. Model conversion, compilation, and deployment often require separate utilities. This slows progress significantly. ModelNova Fusion Studio directly addresses these issues. It offers a unified platform for end-to-end Edge AI development.
embedUR systems announced a major expansion. Fusion Studio now deeply integrates Arm's most adopted Edge AI technologies. This simplifies the entire development lifecycle. It spans from model training and optimization to firmware build and on-device deployment. The announcement came at Embedded World 2026.
The latest Fusion Studio release introduces crucial Arm-native capabilities. Native ExecuTorch support is a key feature. It targets Arm Ethos-U85 and Ethos-U55 NPUs. Models compile via the Vela toolchain. Developers train, quantize, and deploy NPU-optimized models directly. Command-line conversion tools become unnecessary. Framework dependencies disappear.
Arm SDS Framework integration enhances data pipelines. It enables dataset capture, playback, and inference workflows. These operations occur directly from Arm-based hardware targets. Data flows smoothly from sensor to trained model.
Arm Keil MDK toolchain also sees full support. This includes Fixed Virtual Platform (FVP) simulation. Direct Ensemble DevKit deployment is possible. Developers build, test, and iterate efficiently. They remain within the Fusion Studio environment.
Alif Semiconductor Ensemble Series platforms integrate deeply. Full agent and model integration simplifies development. Developers deploy directly to Ensemble Development Kits. Live inference playback is available. Real-time camera capture works on target hardware.
Fusion Studio delivers a complete MLOps pipeline in-tool. It covers dataset capture and annotation. Model training leverages ExecuTorch and Arm Kleidi AI libraries. Firmware builds happen within the application. Direct deployment to Arm-based targets is seamless. All these steps execute locally. This eliminates cloud compute costs.
The new Arm-native capabilities build on a strong foundation. Fusion Studio has been in beta since 2025. It offers a vast library of over 160 pre-trained AI models. These models cover computer vision, audio, and text generation. Twenty-one industry-specific starter packs exist. They accelerate proof-of-concept development.
The platform supports deployment to Raspberry Pi 4 and 5. Integrated camera support is standard. Real-time inference runs efficiently. Fusion Studio operates on various host machines. These include PC (CPU/CUDA) and macOS with Apple Silicon.
Recent beta releases added LLM-powered AI Assistance. This provides contextual guidance. It helps with model selection. Dataset evaluation is simplified. Training configuration becomes easier. Platform optimization is streamlined. Troubleshooting support is readily available. Training checkpoint and continuation support prevent workflow interruptions.
This unified approach brings significant benefits. Fragmented toolchains are a relic of the past. Development cycles shorten dramatically. Innovation accelerates across the board. Teams move from prototype to production faster and more efficiently. The platform removes complexity. It boosts developer productivity.
The ModelNova team continues to invest in expansion. Future Fusion Studio capabilities are in active development. Broader virtual platform support is planned for additional Arm reference designs. GPU server offload will accelerate model training times. Expanded audio capture and annotation workflows are coming. The ability to evaluate models directly on host PC and Apple hardware before deploying to target devices is also on the roadmap. These enhancements are expected throughout 2026.
The latest release, with full Arm ecosystem integration, became available on March 13, 2026. This coincided with the opening of Embedded World 2026. Fusion Studio is available for download at modelnova.com. The ModelNova model zoo, with its extensive library of edge-optimized AI models, can be accessed at modelnova.ai.
embedUR systems demonstrated the full Arm-integrated Fusion Studio workflow at Embedded World 2026. Attendees witnessed ExecuTorch model deployment on Arm Ethos-U NPUs. End-to-end development workflows ran on Ensemble Series hardware. The complete in-tool MLOps pipeline was showcased. A Fusion Studio-built model also ran at the Arm booth. This highlighted the production-readiness of models developed through the platform.
embedUR systems is a Silicon Valley-based embedded software engineering company. It was founded in 2005. The company employs approximately 400 engineers. It operates three core business units: Connectivity, Intelligent Edge, and AI/ML. embedUR has delivered production software globally. Leading semiconductor, consumer electronics, and industrial companies use their solutions.
ModelNova is embedUR's business unit. It offers an open platform. This platform simplifies Edge AI development. It provides pre-trained, optimized models. Curated datasets and deployment blueprints are available. The focus is on resource-constrained edge devices. ModelNova Fusion Studio is its flagship desktop application. It provides a complete end-to-end Edge AI development environment. It eliminates cloud costs. It reduces toolchain complexity. It accelerates time-to-deployment on production hardware.
Edge AI development faces ongoing challenges. Fragmented workflows hinder innovation. Developers assemble many tools. Model conversion, compilation, and deployment often require separate utilities. This slows progress significantly. ModelNova Fusion Studio directly addresses these issues. It offers a unified platform for end-to-end Edge AI development.
embedUR systems announced a major expansion. Fusion Studio now deeply integrates Arm's most adopted Edge AI technologies. This simplifies the entire development lifecycle. It spans from model training and optimization to firmware build and on-device deployment. The announcement came at Embedded World 2026.
The latest Fusion Studio release introduces crucial Arm-native capabilities. Native ExecuTorch support is a key feature. It targets Arm Ethos-U85 and Ethos-U55 NPUs. Models compile via the Vela toolchain. Developers train, quantize, and deploy NPU-optimized models directly. Command-line conversion tools become unnecessary. Framework dependencies disappear.
Arm SDS Framework integration enhances data pipelines. It enables dataset capture, playback, and inference workflows. These operations occur directly from Arm-based hardware targets. Data flows smoothly from sensor to trained model.
Arm Keil MDK toolchain also sees full support. This includes Fixed Virtual Platform (FVP) simulation. Direct Ensemble DevKit deployment is possible. Developers build, test, and iterate efficiently. They remain within the Fusion Studio environment.
Alif Semiconductor Ensemble Series platforms integrate deeply. Full agent and model integration simplifies development. Developers deploy directly to Ensemble Development Kits. Live inference playback is available. Real-time camera capture works on target hardware.
Fusion Studio delivers a complete MLOps pipeline in-tool. It covers dataset capture and annotation. Model training leverages ExecuTorch and Arm Kleidi AI libraries. Firmware builds happen within the application. Direct deployment to Arm-based targets is seamless. All these steps execute locally. This eliminates cloud compute costs.
The new Arm-native capabilities build on a strong foundation. Fusion Studio has been in beta since 2025. It offers a vast library of over 160 pre-trained AI models. These models cover computer vision, audio, and text generation. Twenty-one industry-specific starter packs exist. They accelerate proof-of-concept development.
The platform supports deployment to Raspberry Pi 4 and 5. Integrated camera support is standard. Real-time inference runs efficiently. Fusion Studio operates on various host machines. These include PC (CPU/CUDA) and macOS with Apple Silicon.
Recent beta releases added LLM-powered AI Assistance. This provides contextual guidance. It helps with model selection. Dataset evaluation is simplified. Training configuration becomes easier. Platform optimization is streamlined. Troubleshooting support is readily available. Training checkpoint and continuation support prevent workflow interruptions.
This unified approach brings significant benefits. Fragmented toolchains are a relic of the past. Development cycles shorten dramatically. Innovation accelerates across the board. Teams move from prototype to production faster and more efficiently. The platform removes complexity. It boosts developer productivity.
The ModelNova team continues to invest in expansion. Future Fusion Studio capabilities are in active development. Broader virtual platform support is planned for additional Arm reference designs. GPU server offload will accelerate model training times. Expanded audio capture and annotation workflows are coming. The ability to evaluate models directly on host PC and Apple hardware before deploying to target devices is also on the roadmap. These enhancements are expected throughout 2026.
The latest release, with full Arm ecosystem integration, became available on March 13, 2026. This coincided with the opening of Embedded World 2026. Fusion Studio is available for download at modelnova.com. The ModelNova model zoo, with its extensive library of edge-optimized AI models, can be accessed at modelnova.ai.
embedUR systems demonstrated the full Arm-integrated Fusion Studio workflow at Embedded World 2026. Attendees witnessed ExecuTorch model deployment on Arm Ethos-U NPUs. End-to-end development workflows ran on Ensemble Series hardware. The complete in-tool MLOps pipeline was showcased. A Fusion Studio-built model also ran at the Arm booth. This highlighted the production-readiness of models developed through the platform.
embedUR systems is a Silicon Valley-based embedded software engineering company. It was founded in 2005. The company employs approximately 400 engineers. It operates three core business units: Connectivity, Intelligent Edge, and AI/ML. embedUR has delivered production software globally. Leading semiconductor, consumer electronics, and industrial companies use their solutions.
ModelNova is embedUR's business unit. It offers an open platform. This platform simplifies Edge AI development. It provides pre-trained, optimized models. Curated datasets and deployment blueprints are available. The focus is on resource-constrained edge devices. ModelNova Fusion Studio is its flagship desktop application. It provides a complete end-to-end Edge AI development environment. It eliminates cloud costs. It reduces toolchain complexity. It accelerates time-to-deployment on production hardware.