SiMa.ai | Scaling Physical AI at the Edge

SiMa.ai provides a purpose-built platform for deploying high-performance AI models for vision, audio, and generative applications at the edge with industry-leading power efficiency.

Download our pre-optimized, ready-to-deploy models for the Modalix™ MLSoC below.

• Vision-Language (VLMs): Gemma 3, Qwen2.5-VL, Qwen3-VL, and LFM2-VL.
• Large-Language (LLMs): Llama 3.2/3.1, Phi-3.5-mini, and Mistral-7B, Qwen2.5/3, Gemma
• Audio & Speech: Whisper

Deployment & Tools

• Models are optimized using LLiMa (Automated Code Generation) for sub-10W inference.
• Deployment supports a16w4 and Hybrid a16w8/4 quantization to maximize Performance-per-Watt.
• Join our Developer Portal or visit our Documentation for documentation and SDK access.

Website | GitHub | Contact Support