Deci
Optimize AI model performance and reduce costs with advanced tools.. [Contact for Pricing]
Last verified:
What is Deci?
Deci AI is an end-to-end deep learning acceleration platform that enables AI developers to build, optimize, and deploy faster and more accurate models for any environment including cloud, edge, and mobile. The platform uses AI to build better AI, leveraging Deci's proprietary AutoNAC (Automated Neural Architecture Construction) technology, an algorithmic optimization engine that automatically generates and optimizes deep learning model architectures for specific target hardware while preserving accuracy.
Key features include AutoNAC's Neural Architecture Search component that redesigns trained models to optimize inference performance (throughput, latency, memory), the Infery inference runtime engine for blazing-fast performance with optimal hardware utilization, and SuperGradients, a deep-learning library focused on state-of-the-art computer vision models. The platform applies quantization, pruning, and knowledge distillation techniques to compress models without significant accuracy loss, delivers 3x-15x inference performance improvement on any hardware, and reduces computing costs by up to 80%.
Deci AI is designed for AI developers, data scientists, researchers, and enterprises that need to scale AI initiatives without exponentially increasing cloud infrastructure bills. It serves industries ranging from automotive to smart retail, and is particularly relevant for computer vision, NLP, and Generative AI applications. The platform supports deployment on CPUs, GPUs, FPGAs, and special purpose ASIC accelerators, with special optimization for Intel CPUs achieving record-breaking 11.8x accelerated inference speedup at MLPerf benchmarks.
Deci pricing
Pricing model: Freemium
Deci operates on a Contact for Pricing model with no publicly available pricing information. Pricing is customized based on specific needs, scale of use, and services required such as foundation model access or custom model development. No free tier or free trial is available. The /pricing page contains NVIDIA editorial/marketing content rather than Deci-branded pricing plans. After NVIDIA's acquisition in May 2024, Deci was dissolved as a separate entity and pricing information is accessed through NVIDIA enterprise channels or by contacting NVIDIA sales directly.
Deci pros
- 3x-15x inference performance improvement on any hardware
- AutoNAC automatically optimizes model architecture for target hardware
- Preserves baseline accuracy while accelerating models
- Reduces computing costs by up to 80%
- Supports cloud, edge, and mobile deployment environments
- Hardware-aware optimization for CPU, GPU, FPGA, and ASIC
- Record-breaking 11.8x speedup on Intel CPUs at MLPerf
- Infery SDK delivers blazing-fast inference with minimal latency
- SuperGradients library for state-of-the-art computer vision models
- Quantization, pruning, and knowledge distillation techniques
- Enables new use cases on limited hardware
- Substantially shorter development cycles
- Efficient API for running models across hardware and frameworks
- Visualization tools for analyzing model training metrics
- Generative AI platform with fine-tunable LLMs available
- Virtual Private Cloud or on-premises deployment options
- Named to CBInsights top 100 AI companies
Deci cons
- No public pricing information available - contact for pricing only
- Acquired by NVIDIA in May 2024 and dissolved as independent entity
- Steep learning curve for AutoNAC and neural architecture search
- Primarily focused on enterprise customers, not individual developers
- Custom pricing based on scale makes budgeting difficult
- Documentation scattered across multiple hubs and archives
- No free tier or free trial available
- Requires technical expertise in deep learning optimization
Frequently asked questions about Deci
What is Deci AI?
Deci AI is an end-to-end deep learning acceleration platform that enables AI developers to build, optimize, and deploy faster and more accurate models for any environment including cloud, edge, and mobile. The platform uses AI to build better AI through its proprietary AutoNAC technology that automatically optimizes deep learning model architectures.
What is AutoNAC technology?
AutoNAC (Automated Neural Architecture Construction) is Deci's proprietary algorithmic optimization engine that automatically generates and optimizes deep learning model architectures. It contains a Neural Architecture Search component that redesigns trained models to optimally improve inference performance (throughput, latency, memory) for specific target hardware while preserving baseline accuracy, providing up to 10X performance boost and up to 80% cost savings.
Who can use Deci AI?
Deci AI is designed for AI developers, data scientists, researchers, and enterprises that need to scale their AI initiatives. It is particularly suitable for teams working on computer vision, NLP, and Generative AI applications, and for industries ranging from automotive to smart retail that need to deploy efficient deep learning models on various hardware.
What hardware does Deci support?
Deci's platform is hardware-aware and optimizes models for CPUs, GPUs, FPGAs, and special purpose ASIC accelerators. The platform works across cloud, edge, and mobile environments, with special optimization capabilities for Intel CPUs where it achieved record-breaking performance at MLPerf benchmarks.
What is Infery?
Infery is Deci's inference runtime engine that delivers blazing-fast performance with optimal hardware utilization. It is designed to be highly efficient and scalable, allowing users to run deep learning models with minimal latency and maximum throughput. Infery applies techniques like quantization, pruning, and knowledge distillation and is particularly well-suited for real-time decision-making applications like autonomous vehicles, robotics, and medical imaging.
What is SuperGradients?
SuperGradients is Deci's powerful deep-learning library specifically focused on state-of-the-art models of computer vision. It provides tools and resources to optimize deep learning models, resulting in faster training times and better overall performance. It is a popular choice among developers and researchers working on cutting-edge computer vision projects and is the home of Yolo-NAS.
Is Deci AI easy to learn?
Deci AI requires technical expertise in deep learning optimization. The platform offers an efficient API that is easy to use and scalable for deploying models, but working with AutoNAC and neural architecture search requires understanding of machine learning concepts. Documentation is available through the Deci AI Documentation Hub covering SuperGradients, Infery, and the Deep Learning Platform.
What types of AI models does Deci optimize?
Deci specializes in optimizing models for computer vision and language domains, including Generative AI. The platform can optimize any deep neural solution, with particular strength in computer vision models through SuperGradients. The Generative AI Platform features proprietary fine-tunable large language models (LLMs) along with an inference engine and AI inference cluster management solution.
How does Deci reduce costs?
Deci reduces computing costs by up to 80% through its AutoNAC technology that optimizes model architecture for maximum hardware utilization. By accelerating inference performance 3x-15x while preserving accuracy, companies can perform more deep learning tasks in real-time on edge devices and dramatically cut cloud or datacenter costs, even by changing inference hardware from GPU to more cost-effective Intel CPUs.
What happened to Deci after NVIDIA acquisition?
Deci was acquired by NVIDIA Corporation in May 2024 and was dissolved as an independent corporate entity. The deci.ai domain now serves NVIDIA-branded content including newsroom-style content about NVIDIA AI computing leadership. Pricing information is no longer available on Deci's website and must be accessed through NVIDIA enterprise channels or by contacting NVIDIA sales directly.