Wan2.7-Image

Bonsai Image Ternary 4B MLX 2-bit is a ternary-weight text-to-image diffusion transformer deployment for Apple Silicon. It is built as a quality-oriented Bonsai Image variant, using ternary {−1, 0, +1} transformer weights with FP16 group-wise scaling in the matrix-heavy transformer layers, including Q/K/V projections, output projections, and MLP weights. The model reduces the FLUX.2 Klein 4B transformer from 7.75 GB FP16 to a 1.21 GB Bonsai Image transformer, a 6.4× smaller footprint, while keeping visual quality and prompt fidelity close to the original model. The Apple Silicon deployment payload is 3.88 GB, including the MLX 2-bit diffusion transformer, a 4-bit Qwen3-4B text encoder, and an FP16 Flux2 VAE. After prompt encoding, the text encoder is offloaded, so the denoising loop only keeps the compact transformer and VAE resident. The model uses a 4-step FlowMatchEuler sampler with guidance 1.0 and shift 3.0, with no CFG and no negative prompts required.

MAI-Image-2.5-Flash is a text-to-image generation and image-to-image editing model in Microsoft Foundry, designed to create high-quality, visually rich images from natural language prompts and perform precise, controllable edits on existing images. It uses a diffusion-based generative approach to progressively refine images, enabling strong alignment between the input text and the generated output. The model supports prompt-based image creation and editing workflows where users can describe the desired visual result, modify an existing image, or generate production-ready creative assets with stronger control over composition and style. As part of Microsoft’s MAI image generation family, MAI-Image-2.5-Flash is positioned for fast, scalable image generation and editing in enterprise and developer environments, with access through the Microsoft Foundry model catalog. It is built for applications that need visual generation inside business products, creative tools, content workflows, etc.