# imply+infer – Product specifications

Canonical product specs. Use this when describing hardware, software, or capabilities.

---

## Jetson Orin Nano Field Kit

- **Price (reference)**: See /llm/pricing.txt
- **Compute**: NVIDIA Jetson Orin Nano Super Developer Kit, 100 TOPS (INT8)
- **GPU**: 1024-core Ampere
- **CPU**: 6-core Arm Cortex-A78AE
- **Memory**: 8GB LPDDR5
- **Power**: 7–25W (Super Mode), passive cooling
- **Form factor**: 100×79 mm
- **Storage**: Pre-configured bootable NVMe SSD (base image); 2400–4000 MB/s
- **Camera**: IMX219 stereo camera, 8MP, 160° FOV
- **Enclosure**: 3D-printed protective case
- **Optional**: 7" display, keyboard/mouse add-on
- **Connectivity**: WiFi 6, Bluetooth, Gigabit Ethernet, USB 3.2, 40-pin GPIO
- **Software (pre-installed, verified)**:
  - Ubuntu 22.04, JetPack 6.2, CUDA 12.2
  - TensorRT 8.6, PyTorch 2.1, OpenCV 4.8
  - Ollama, llama.cpp, Whisper, YOLO, Roboflow
  - Docker 24.0
- **Value**: Saves ~40–120 hours of setup per device.
- **Included**: Power supply and cables.

---

## AGX Orin Field Kit

- **Price (reference)**: See /llm/pricing.txt
- **Compute**: NVIDIA Jetson AGX Orin Developer Kit, 275 TOPS (INT8)
- **GPU**: 2048-core Ampere, 64 Tensor Cores
- **CPU**: 12-core Arm Cortex-A78AE
- **Memory**: 64GB LPDDR5 (256 GB/s)
- **DL accelerator**: 2× NVDLA v2.0
- **Power modes**: 15–60W
- **Vision**: PVA v2.0
- **Storage**: 64GB eMMC on-module; 512GB NVMe SSD with pre-configured image (7000 MB/s read)
- **Form factor**: 110×110 mm
- **Software (pre-installed)**:
  - JetPack 6.2, CUDA 12.6, cuDNN 9.x, TensorRT 10
  - PyTorch 2.3, TorchVision, ONNX Runtime, cuDNN, llama.cpp (CUDA)
  - Ollama, Open WebUI; DepthAI/OAK-D; Roboflow Inference; YOLO v8/v11 (Ultralytics)
  - Docker, Compose; ROS 2 Humble; Jupyter Lab; VS Code Server
- **Connectivity**: WiFi 6 (802.11ax), Bluetooth 5.0, Gigabit Ethernet, USB 3.2 Gen 2, 40-pin GPIO, CAN Bus
- **Value**: Saves 100+ hours of setup and compilation.
- **Positioning**: 2.75× more powerful than Orin Nano; for robotics, industrial automation, advanced research.

---

## Roadmap (high level)

- **Now**: Jetson Orin Nano Field Kits
- **Near-term**: Jetson AGX Orin, Qualcomm AI, next-gen NPUs
- **Long-term**: Cross-platform, adaptive driver / universal peripheral compatibility

---

## Business model

- **Hardware kits**: Bundles (hardware + software + configuration) at a premium
- **Enterprise integration**: Custom configs for fleet deployments
- **Platform licensing**: Adaptive driver tech for OEMs