simaai/gemma3-siglip448-a16w4
Image-Text-to-Text
•
Updated
•
175
AI & ML interests
None defined yet.
Recent Activity
Organization Card
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.
- Generative AI (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.
models
24
simaai/Qwen2.5-VL-3B-Instruct-a16w4
Image-Text-to-Text
•
Updated
simaai/LFM2-VL-3B-a16w4
Image-Text-to-Text
•
Updated
simaai/LFM2-VL-1.6B-a16w4
Image-Text-to-Text
•
Updated
simaai/LFM2-VL-450M-a16w4
Image-Text-to-Text
•
Updated
simaai/Qwen3-VL-4B-Instruct-a16w4
Image-Text-to-Text
•
Updated
simaai/Qwen3-VL-8B-Instruct-a16w4
Image-Text-to-Text
•
Updated
simaai/Phi-3.5-mini-instruct-a16w4
Text Generation
•
Updated
•
38
simaai/Mistral-7B-Instruct-v0.3-a16w4
Text Generation
•
Updated
•
38
simaai/Llama-3.2-3B-Instruct-a16w4
Text Generation
•
Updated
•
128
datasets
0
None public yet