A curated collection of ready-to-use AI models compatible with the blace.ai inference SDK.
Segment Anything 3 — Image Segmentation 👉 Model Card
Depth Anything 2 — Depth Estimation 👉 Model Card
Memfof — Optical Flow 👉 Model Card
| Model | Configuration | Category | Link |
|---|---|---|---|
| Distillanydepth | default | Depth Estimation | Model Card |
| Distillanydepth | small | Depth Estimation | Model Card |
| Distillanydepth | large | Depth Estimation | Model Card |
| BEN2 | default | Human Matting | Model Card |
| Coordfill | default | Image Inpainting | Model Card |
| Depth Anything V2 | small | Depth Estimation | Model Card |
| Depth Anything 3 | metric_large | Depth Estimation | Model Card |
| Depth Anything 3 | mono_large | Depth Estimation | Model Card |
| Gemma | 2b-v2 | LLM | Model Card |
| Llmdet | default | Open-Vocabulary Object Detection | Model Card |
| Memfof | default | Optical Flow Estimation | Model Card |
| Raft | default | Optical Flow Estimation | Model Card |
| Retinexformer | default | Low Light Image Enhancement | Model Card |
| Segment Anything 3 | default | Image Segmentation | Model Card |
| Semantic Guided Human Matting | default | Human Matting | Model Card |
| Videoflow | default | Optical Flow Estimation | Model Card |
Our hub is designed for developers who want cross-platform, plug-and-play AI in their native C++ applications with minimal setup. Model features:
- ✅ Can be included with a few lines of code
- ✅ Continuously tested across supported devices and backends
- ✅ Self-contained — no Python required
- ✅ Ship with easy-to-use demos
- ✅ Hardware-Accelerated on CUDA, MPS and DirectML
You can make your custom Torchscript and ONNX models compatible with the blace.ai runtime (Documentation).
A searchable database of all models is available under https://blace.ai/hub.