stable-diffusion-v1-5

Stable Diffusion v1.5 (≈ 1 B params, CreativeML Open RAIL-M)

Text-to-image generator that fits on a single consumer GPU.

• What it is. Latent-diffusion model that turns plain-English prompts into 512 × 512 images; upscale or edit after if you need bigger.

• Under the hood. ~860 M-parameter U-Net + 123 M CLIP ViT-L/14 text encoder run in latent space—lighter and faster than pixel-space diffusion.

• Runs on cheap hardware. Needs about 5-8 GB VRAM for FP16 inference; 4-bit or CPU builds will run on 4 GB cards or laptops, just slower..

• Drop-in tooling. One-liner with diffusers, plus full support in Automatic1111, ComfyUI, SD.Next, InvokeAI, etc.—same workflow as any SD-1.x model.

• License quirks. CreativeML Open RAIL-M lets you use it commercially but bans porn, extreme gore, hateful or copyright-breaking content.

Why pick it for Norman AI?

We can auto-generate hero art, social banners, or quick mock-ups on a single A10G or even a beefy MacBook. No vendor lock-in, no extra infra—just load the weights, point our brand-prompt template at it, and ship visuals fast.

messages = [
    {"role": "system", "content": "You are a helpful AI assistant."},
    {"role": "user", "content": "Can you provide ways to eat combinations of bananas and dragonfruits?"},
    {"role": "assistant",
     "content": "Sure! Here are some ways to eat bananas and dragonfruits together"},
    {"role": "user", "content": "What about solving an 2x + 3 = 7 equation?"},
]

response = await norman.invoke(
    {
        "model_name": "phi-4",
        "inputs": [
            {
                "display_title": "Prompt",
                "data": messages
            }
        ]
    }
)