The SiMa.ai MLSoC Modalix Development Kit (DevKit) is a complete hardware and software platform for evaluating and prototyping intelligent Physical AI applications using the latest Modalix System-on-Module (SoM).
Designed for rapid deployment, this kit enables seamless development, benchmarking, and proof-of-concept creation with SiMa.ai’s Palette software, offering best-in-class performance-per-watt in a compact form factor. Ideal for industrial, robotics, and vision-based and LLM-based applications, this kit supports real-time inferencing across computer vision, transformer models, and GenAI workloads.
Launch edge AI applications in just minutes with our streamlined workflow
Design Physical AI, robotics, and sensor-based apps with ease
Access ready-to-run examples, pretrained models, and full guides
Fast deployment includes software, cables, tools, and power—no extras needed
Deploy your models seamlessly with industry-standard compatibility
The MLSoC Modalix System-on-Module (SoM) provides the proven performance of the Modalix MLSoC architecture in a compact, deployment-ready form factor, purpose-built to Scale Physical AI applications.
The Modalix innovative system architecture seamlessly integrates a set of compute engines and peripherals, delivering best-in-class performance per watt for advanced workloads, including multimodal Transformers, Large Language Models (LLMs), Large Multimodal Models (LMMs), and Generative AI (GenAI). It also maintains full support for legacy convolutional neural networks (CNNs) and traditional
computer vision algorithms, ensuring broad compatibility across both modern and established ML pipelines.
The Modalix SoM offers the best-in-class performance per watt.
| Machine Learning Accelerator (MLA) | 50 TOPs (BF16, INT8, INT16) Supports LLMs, neural networks, and GenAI |
| Application Compute Unit (ACU) | 8 × ARM Cortex-A65 @ 1.4 GHz 32K Dhrystone MIPS |
| Memory | 32 GB or 8 GB on-board 128-bit LPDDR5 (6400 Mbps) system memory |
| Boot and Security Unit (BSU) | Secure boot with boot code authentication and encryption Secure key storage and management Security engine for user code |
| Video Codec | Decode H.264 / H.265 / AV1 up to 4K60 Encode H.264 / H.265 up to 4K60 |
| Computer Vision Unit (CVU) | 4-core Synopsys EV74 DSP @ 1 GHz 720 GOPS (16-bit) |
| Low-Latency ISP | ARM Mali-C-1 @ 1.2 GHz |
| Display Connectivity | HDMI 1.4 port with 4K resolution |
| Peripherals | 4 × 2-lane MIPI CSI-2 4 × PCIe Gen5 RC & EP 1 × 1Gb Ethernet PHY 3 × USB 3.0 4 × I2C, 3 × UART, 14 × GPIO 2 × SPI |
| Storage | External NVMe via PCIe x4 External SSD via USB 3.0 16 GB eMMC flash 64 MB QSPI flash 128 KB EEPROM |
| Power Supply | Flexible 5–20 V power input |
| Form Factor | 69.6 mm × 45 mm 260-pin SO-DIMM connector |
| Operating Temperature |
Commercial SoM: 0°C to +60°C Industrial SoM: -40°C to +85°C |
| Phase | Description | Details |
|---|---|---|
| Evaluate | Use Palette to compile and run models, assess performance (FPS, latency, compute utilization), and monitor system metrics. |
|
| Prototype | Developers can rapidly prototype functional pipelines using SiMa.ai’s Python APIs. | Enables fast bring-up of custom ML workflows directly on the hardware without complex embedded optimization. |
| Demonstrate | The Deployment supports multi-camera input and high-speed interfaces for showcasing applications. |
|
| No. | Interfaces |
|---|---|
| 1 | HDMI 1.4 @ 4K |
| 2 | 2 × MIPI CSI-2 Connectors |
| 3 | M.2 M — 4× PCIe Gen4 RC/EP |
| 4 | 4× USB 3.2 |
| 5 | 1× USB-C |
| 6 | 1× Gigabit Ethernet |
| 7 | I²C / UART / GPIO / SPI / GPIO |
| Product Brief | Download |
| Parts | Quantity |
|---|---|
| Modalix SoM 32GB (Commercial/Industrial) | 1 |
| Carrier Board | 1 |
| Active Heatsink | 1 |
| M.2 240GB NVME Solid State Drive | 1 |
| 12V/5A Power Supply | 1 |
| Metal Enclosure | 1 |
No reviews yet. Be the first to share your experience!
For more discussion Click Here
Discover More In:
Category: View all Barebone Computers
Fill out the form below and our team will get back to you with bulk pricing.
The SiMa.ai MLSoC Modalix Development Kit (DevKit) is a complete hardware and software platform for evaluating and prototyping intelligent Physical AI applications using the latest Modalix System-on-Module (SoM).
Designed for rapid deployment, this kit enables seamless development, benchmarking, and proof-of-concept creation with SiMa.ai’s Palette software, offering best-in-class performance-per-watt in a compact form factor. Ideal for industrial, robotics, and vision-based and LLM-based applications, this kit supports real-time inferencing across computer vision, transformer models, and GenAI workloads.
Launch edge AI applications in just minutes with our streamlined workflow
Design Physical AI, robotics, and sensor-based apps with ease
Access ready-to-run examples, pretrained models, and full guides
Fast deployment includes software, cables, tools, and power—no extras needed
Deploy your models seamlessly with industry-standard compatibility
The MLSoC Modalix System-on-Module (SoM) provides the proven performance of the Modalix MLSoC architecture in a compact, deployment-ready form factor, purpose-built to Scale Physical AI applications.
The Modalix innovative system architecture seamlessly integrates a set of compute engines and peripherals, delivering best-in-class performance per watt for advanced workloads, including multimodal Transformers, Large Language Models (LLMs), Large Multimodal Models (LMMs), and Generative AI (GenAI). It also maintains full support for legacy convolutional neural networks (CNNs) and traditional
computer vision algorithms, ensuring broad compatibility across both modern and established ML pipelines.
The Modalix SoM offers the best-in-class performance per watt.
| Machine Learning Accelerator (MLA) | 50 TOPs (BF16, INT8, INT16) Supports LLMs, neural networks, and GenAI |
| Application Compute Unit (ACU) | 8 × ARM Cortex-A65 @ 1.4 GHz 32K Dhrystone MIPS |
| Memory | 32 GB or 8 GB on-board 128-bit LPDDR5 (6400 Mbps) system memory |
| Boot and Security Unit (BSU) | Secure boot with boot code authentication and encryption Secure key storage and management Security engine for user code |
| Video Codec | Decode H.264 / H.265 / AV1 up to 4K60 Encode H.264 / H.265 up to 4K60 |
| Computer Vision Unit (CVU) | 4-core Synopsys EV74 DSP @ 1 GHz 720 GOPS (16-bit) |
| Low-Latency ISP | ARM Mali-C-1 @ 1.2 GHz |
| Display Connectivity | HDMI 1.4 port with 4K resolution |
| Peripherals | 4 × 2-lane MIPI CSI-2 4 × PCIe Gen5 RC & EP 1 × 1Gb Ethernet PHY 3 × USB 3.0 4 × I2C, 3 × UART, 14 × GPIO 2 × SPI |
| Storage | External NVMe via PCIe x4 External SSD via USB 3.0 16 GB eMMC flash 64 MB QSPI flash 128 KB EEPROM |
| Power Supply | Flexible 5–20 V power input |
| Form Factor | 69.6 mm × 45 mm 260-pin SO-DIMM connector |
| Operating Temperature |
Commercial SoM: 0°C to +60°C Industrial SoM: -40°C to +85°C |
| Phase | Description | Details |
|---|---|---|
| Evaluate | Use Palette to compile and run models, assess performance (FPS, latency, compute utilization), and monitor system metrics. |
|
| Prototype | Developers can rapidly prototype functional pipelines using SiMa.ai’s Python APIs. | Enables fast bring-up of custom ML workflows directly on the hardware without complex embedded optimization. |
| Demonstrate | The Deployment supports multi-camera input and high-speed interfaces for showcasing applications. |
|
| No. | Interfaces |
|---|---|
| 1 | HDMI 1.4 @ 4K |
| 2 | 2 × MIPI CSI-2 Connectors |
| 3 | M.2 M — 4× PCIe Gen4 RC/EP |
| 4 | 4× USB 3.2 |
| 5 | 1× USB-C |
| 6 | 1× Gigabit Ethernet |
| 7 | I²C / UART / GPIO / SPI / GPIO |
| Product Brief | Download |
| Parts | Quantity |
|---|---|
| Modalix SoM 32GB (Commercial/Industrial) | 1 |
| Carrier Board | 1 |
| Active Heatsink | 1 |
| M.2 240GB NVME Solid State Drive | 1 |
| 12V/5A Power Supply | 1 |
| Metal Enclosure | 1 |
No reviews yet. Be the first to share your experience!
For more discussion Click Here
Discover More In:
Category: View all Barebone Computers
Fill out the form below and our team will get back to you with bulk pricing.
Custom Requirements?
Related products
