The Raspberry Pi It has become much more than just a simple board for tinkering at home. In recent months, new products and solutions based on this platform have emerged, directly targeting edge artificial intelligence, reliable storage, and professional embedded systems, with a particular impact on the European and Spanish markets.
From a new official flash drive Designed to withstand demanding uses, from a surveillance camera with integrated artificial vision to an AI HAT specifically designed for generative models on Raspberry Pi 5, the brand is reinforcing an ecosystem that is no longer limited to the educational or maker field, but is starting to compete head-to-head in industrial, security and data analysis projects without depending on the cloud.
AI HAT+ 2: Generative Artificial Intelligence and Vision on Raspberry Pi 5
The launch of the AI HAT+ 2 This marks a significant shift in how Raspberry Pi approaches artificial intelligence. Until now, much of the focus was on classic computer vision tasks (object detection, segmentation, etc.), but the rise of generative language models The landscape has changed: it is no longer enough to "see", now devices have to "understand and respond".
This new expansion board is designed for the Raspberry Pi 5 and integrates a Hailo-10H accelerator capable of delivering up to 40 TOPS (INT4), accompanied by 8 GB of dedicated RAM. This combination allows for handling larger and more complex models directly on the device, without resorting to external servers, which is very useful in environments where the data privacy and low latency They are critical.
One of the key features of the AI ​​HAT+ 2 is that it maintains compatibility with the vision workflows already known in the Raspberry Pi ecosystem, so users who worked with models like YOLO or with segmentation algorithms can continue to take advantage of that capability, while now adding generative AI functions in the same hardware.
Raspberry Pi has confirmed that, from launch, it will offer support for edge-optimized language models, such as DeepSeek-R1-Distill o llama 3.2Although they are more compact than their cloud-based equivalents, they can be adapted using fine-tuning techniques such as LoRA and the use of the Hailo Dataflow Compiler, which opens the door to specialized assistants: from local translation to description of scenes captured in real time by cameras connected to the board itself.
In terms of integration, this HAT maintains compatibility with the official camera ecosystem from Raspberry Pi, facilitating the upgrade of existing projects that already use computer vision. Furthermore, the company has indicated that the product will have a launch price of around $130 and its production is guaranteed at least until 2036This is a very relevant piece of information for European companies that need supply stability for their long-term designs.
With this move, Raspberry Pi strengthens its position in sectors such as robotics, offline data analysis, or process control in environments without a reliable internet connection, where relying on cloud services is not an option or could make the project too expensive.
Raspberry Pi Flash Drive: the official flash drive designed to last
Another of the notable new features within the ecosystem is the Raspberry Pi Flash Drive, an official USB 3.0 flash drive available in capacities of 128 and 256 GBIt may seem like just another flash drive in a market full of cheap options, but here the focus is not only on the price, but on the combination of speed, stability and resistance to rough use.
While many low-cost USB drives offer inconsistent speeds, lower actual capacity than advertised, or failures in the event of power outages, Raspberry Pi has focused on the long term reliabilityThe device incorporates a durable aluminum casing, with a small hole for attaching it to a keychain or backpack, and connects via USB-A 3.0This makes it compatible with both minicomputers and conventional PCs.
The version of 128 GB achieves approximately 75 MB/s of sustained sequential write speed, while the model of 256 GB reaches approximately 150 MB/sThis performance jump between one model and another can make all the difference if you plan to use it as a drive for continuous data recording or even for run operating systems and Raspberry Pi distributions directly from USB.
Internally, the unit resorts to a pseudo-SLC cache which acts as a fast zone to absorb the most intense write bursts. In the background, this information is transferred to higher-density but slower QLC flash memory. This approach allows write speeds to approach the practical limit of the USB 3.0, without performance plummeting when the load is maintained over time.
Another key point is robustness against unexpected withdrawals or power outagesAccording to data provided by Raspberry Pi itself, the unit has been subjected to tens of thousands of random power cycles under intensive input/output workloads, precisely to validate that data corruption does not occur easily.
The Raspberry Pi Flash Drive also includes support for SMART reports and TRIM operationsThese functions are more common in SSDs than in simple USB drives. This allows you to monitor the memory's status and maintain its performance throughout the device's lifespan. Furthermore, it automatically enters different states of low power USB 3.0 when it is idle, which is relevant in battery-powered projects or installations with many connected units.
In Spain, this USB drive is already available for purchase at stores specializing in Raspberry Pi such as TiendaTec, Rubii, or RaspiPC. The approximate price is around 29,99 euros for the 128 GB version and around 55,95 euros for the 256 GBThese figures are somewhat higher than those of generic models, but justified by the emphasis on stability and durability for more serious uses than simply transporting files.
Smart cameras based on Raspberry Pi Compute Module 5 and Sony IMX500 sensor
Alongside the development of new HATs and storage solutions, they are arriving on the European market advanced surveillance cameras that use the Raspberry Pi platform as the heart of the system. A prime example is an outdoor camera that combines a Raspberry Pi Compute Module 5 (CM5) with the machine vision sensor Sony IMX500 12,3 megapixels.
The use of Compute Module 5 responds to the need for a industrial and compact formatdesigned to be integrated into final products rather than prototypes. In this case, the typical configuration includes 4 GB of RAM and 16 GB of eMMC storagesufficient to run a complete Linux system, video capture services, network communications, and the application logic required for remote management and local analysis.
The differentiating factor is the sensor. Sony IMX500, capable not only of capturing images at resolutions up to 4056 × 3040 pixelsbut also to run models of machine learning directly on the sensor itself thanks to its integrated inference accelerator. This architecture, already documented in the Raspberry Pi AI Camera, allows generating processed outputs (such as detections or classifications) without saturating the main CPU of the CM5.
From the perspective of edge vision, these types of devices significantly reduce both the latency such as bandwidth requiredInstead of continuously sending video streams to the cloud, the camera processes scenes locally and only transmits relevant results: detection of people, vehicles, intrusions, or object counts in industrial environments.
The IMX500 sensor can work at its maximum resolution at about 10 frames per second, or operate at resolutions of approximately 2 megapixels at 30 fpsThis provides the flexibility to adjust the balance between detail and performance depending on the application. This approach is well-suited to perimeter surveillance, access control, or process monitoring in warehouses and factories.
The camera is integrated into a Metal housing with IP66 certificationThis ensures protection against dust and water jets, essential for permanent outdoor installations. The recommended operating range is approximately between 0 and 50 ° C, sufficient for most European climates, although in extreme environments additional protection or air conditioning systems may be required.
For connectivity and power supply, the following has been chosen: Ethernet with PoE Under the IEEE 802.3at standard, a single cable provides both data and power. Power consumption is around 15 WIn line with other professional IP cameras, this solution simplifies installation, reduces points of failure, and facilitates maintenance—features valued by integrators and security companies.
What makes this type of camera especially interesting compared to homemade setups with standard Raspberry Pi and generic camera modules is its approach as an integrated productWe are talking about a closed, robust solution designed to work continuously, which can serve as a basis for European developers to build intelligent surveillance systems, traffic analysis or advanced monitoring without starting from scratch in hardware.
With prices around $380These devices occupy a middle ground between fully enclosed IP cameras and custom-built solutions. Specialized embedded hardware distributors within the EU, such as Fab to Lab and similar retailers, are presenting them as kits geared towards professional applicationswhere the flexibility of the Raspberry Pi ecosystem scores points against less modifiable devices.
The set of launches and products based on Raspberry Pi This demonstrates that the platform is taking a firm step towards a more professional landscape: from the AI ​​HAT+ 2 focused on generative AI and computer vision, to the official flash drive designed to offer stable and long-lasting storage, to industrial cameras with Compute Module 5 and sensors capable of processing models directly on the chip. All of this positions Raspberry Pi as a highly competitive option for projects in Spain and Europe that require edge computing, privacy, low power consumption, and a mature hardware and software ecosystem without relying exclusively on cloud services.
