Geniatech AIM-M-K is a M.2 2280 AI accelerator powered by the NXP Ara-240 NPU, delivering up to 40 TOPS. Using PCIe Gen4 over M.2 M-key, it enables edge systems to run transformer and generative AI models locally with low latency and no cloud dependency – all without major hardware redesign on existing x86 or Arm-based platforms.

• Up to 40 TOPS AI performance for next-gen inference (NXP Ara-240 NPU)
• Generative AI at the edge – CNNs, LLMs, transformers & multimodal models
• Multi-model concurrency – concurrent multi-stream inferencing
• Standard M.2 2280 (M-key) – drop-in AI for industrial & embedded platforms
• PCIe Gen4 x4 interface for high data throughput
• Up to 16 GB LPDDR4 on-board memory for complex AI models
• Framework support – TensorFlow, PyTorch, ONNX
• Low latency – ResNet50 @ 2ms | Stable Diffusion 1.4 @ 7 sec/image