Every model, by platform.
Pick your hardware, then follow the guide for the model you want to run. Weights are supplied by you from their publishers; Veizik ships only the engine.
NVIDIA · CUDA
Qwen language models and Wan video models on your own NVIDIA GPU. Available now: Qwen3.5 · Qwen3.6 · Wan 2.1 · Wan 2.2. Not in a download yet: FLUX.1 · Stable Diffusion 3.5.
Run the Qwen3.5 dense checkpoints in veizik's native runtime. Weights are user-supplied, not shipped.
Run the Qwen3.6-35B-A3B mixture-of-experts checkpoint in veizik's native runtime; on NVIDIA it also takes an image or a short video with the prompt. Weights are user-supplied, not shipped.
Run the Wan2.1 and Wan2.2 text-to-video and image-to-video checkpoints in veizik's native runtime. Weights are user-supplied, not shipped.
Still-image generation from text, run directly from the published checkpoints.
Still-image generation from text, read straight from the published safetensors checkpoints.
Stable Diffusion 3.5: Powered by Stability AI
Apple Silicon · Metal
Available now. Run these today on any Apple Silicon Mac.
Run Qwen2.5 directly from supported safetensors checkpoints with no converted model copy.
Run the post-trained Qwen3-1.7B checkpoint in a compact native runtime. Qwen3-1.7B-Base is catalogued but not published.
Raspberry Pi · edge
Generative AI on a Raspberry Pi CPU — no GPU, no Python. The edge engine is in preparation; the Edge page has the on-device measurements and samples.
Run Stable Diffusion 1.5 in veizik's native edge runtime on a Raspberry Pi CPU — no GPU, no Python. Weights are user-supplied from the publisher, not shipped.
Run SDXS in veizik's native edge runtime on a Raspberry Pi CPU — no GPU, no Python. Weights are user-supplied from the publisher, not shipped.
Run SmolLM2 in veizik's native edge runtime on a Raspberry Pi CPU — no GPU, no Python. Weights are user-supplied from the publisher, not shipped.
Run GPT-2 in veizik's native edge runtime on a Raspberry Pi CPU — no GPU, no Python. Weights are user-supplied from the publisher, not shipped.
Run LFM2.5 in veizik's native edge runtime on a Raspberry Pi CPU — no GPU, no Python. Weights are user-supplied from the publisher, not shipped.
Run NVIDIA's SANA diffusion models in veizik's native edge runtime on a Raspberry Pi CPU. Weights are user-supplied from the publisher, not shipped.
All 16 families.
| Family | Class | Publisher | Variants | Platform | Status | Build |
|---|---|---|---|---|---|---|
| Qwen2.5 | Language | Alibaba (Qwen) | 6 | Apple Silicon | Available | 2026.10.01 |
| Qwen3 | Language | Alibaba (Qwen) | 2 | Apple Silicon | Available | 2026.10.01 |
| Qwen3.5 | Language | Alibaba (Qwen) | 4 | NVIDIA | Available | cuda-qwen-1.1.4 |
| Qwen3.6 | Language | Alibaba (Qwen) | 1 | NVIDIA | Available | cuda-qwen-1.1.4 |
| Wan 2.1 · Wan 2.2 | Video | Wan-AI (Alibaba) | 4 | NVIDIA | Available | cuda-wan-1.2.92 |
| FLUX.1 | Image | Black Forest Labs | 8 | NVIDIA | Coming next | — |
| Stable Diffusion 3.5 | Image | Stability AI | 2 | NVIDIA | Coming | — |
| FLUX.2 | Image | Black Forest Labs | 1 | — | Compatible | — |
| HunyuanImage | Image | Tencent | 2 | — | Compatible | — |
| HunyuanVideo | Video | Tencent | 2 | — | Compatible | — |
| Stable Diffusion 1.5 | Image | Stability AI · RunwayML | 1 | Raspberry Pi · edge | Coming | — |
| SDXS | Image | IDKiro (SDXS) | 1 | Raspberry Pi · edge | Coming | — |
| SmolLM2 | Language | Hugging Face | 2 | Raspberry Pi · edge | Coming | — |
| GPT-2 | Language | OpenAI | 1 | Raspberry Pi · edge | Coming | — |
| LFM2.5 | Language | Liquid AI | 1 | Raspberry Pi · edge | Coming | — |
| SANA | Image | NVIDIA | 4 | Raspberry Pi · edge | Coming | — |
“Compatible” means the engine runs the architecture; it is not offered as a download yet. Build is the release each supported model shipped in.
Stable Diffusion 3.5: Powered by Stability AI HunyuanImage: Weights are not included. Model availability and use are governed by the model provider's license and regional restrictions.