Creative Lite
Turn a short prompt into a polished image with a private, on-device workflow.
FREE AND OPTIONAL CLOUD MODELS
Free AI image models and paid cloud options.
Offlair includes free image models and offers paid cloud models when you want more choices.
MODEL CATALOG
Two verified candidates. Eight research directions.
Creative Lite and Enhance Lite passed current physical-device testing. Other modes remain internal research candidates.
Enhance Lite
Add clarity and scale a supported creation into a more detailed image.
Photo Pro
Shape lifelike scenes with balanced light, texture, and realistic depth.
Creative Pro
Build striking concepts with rich color, atmosphere, and confident composition.
Anime Lite
Create bright illustrated characters, playful scenes, and clean animated styles.
Photo Ultra
Explore premium photographic detail for portraits, products, and cinematic scenes.
Anime Pro
Develop refined characters and layered illustrated worlds with a distinct visual voice.
Edit Ultra
Rework an existing image while keeping the parts that matter most to you.
Design Ultra
Turn clear ideas into bold visual concepts for posters, covers, and social posts.
Creative Ultra
Push ambitious prompts toward artful, memorable images with a distinctive finish.
MODEL GUIDES
Understand each premium image model.
Review capabilities, controls, launch timing, best uses, and limits before choosing a cloud model.
Google DeepMind
Nano Banana 2
A fast image generation and conversational editing model based on Gemini 3.1 Flash Image.
Read model guideGoogle DeepMind
Nano Banana Pro
A professional image generation and editing model based on Gemini 3 Pro Image.
Read model guideOpenAI
GPT Image 2
An image generation and editing model built for detailed instructions, typography, and high-fidelity image inputs.
Read model guideBlack Forest Labs
FLUX 2 Max
A high-quality FLUX.2 model for precise image generation, realism, and visual consistency.
Read model guideTopaz Labs
Topaz Upscale
An image enhancement service for upscaling, detail recovery, denoising, sharpening, and face recovery.
Read model guidexAI
Grok Imagine Image
A cloud image generation model for fast, aesthetic image creation from text prompts.
Read model guideBlack Forest Labs
FLUX.2 Turbo
A speed-focused hosted FLUX.2 variant for rapid image generation and prompt iteration.
Read model guideIdeogram
Ideogram V3
A hosted image model with typography, reference, palette, and design controls.
Read model guideBlack Forest Labs
FLUX.1 Kontext Pro
A previous-generation hosted model for prompt-based generation and contextual image editing.
Read model guideByteDance Seed
Seedream 4.5
A hosted generation and editing model focused on reference consistency and dense text.
Read model guideAlibaba Qwen
Qwen Image
A 20B image foundation model for text rendering, generation, and precise editing.
Read model guideTencent Hunyuan
Hunyuan Image 3.0
An open native multimodal text-to-image model with 80B total parameters and 13B active parameters.
Read model guideTencent Hunyuan
Hunyuan Image 3.0 Instruct
An instruction variant that supports text-to-image and text-plus-image transformation.
Read model guideBlack Forest Labs
FLUX.1 Dev
An open-weight FLUX.1 model for development, research, and customizable image generation.
Read model guideBlack Forest Labs
FLUX.1 Schnell
A step-distilled open-weight FLUX.1 model designed for one-to-four-step generation.
Read model guideRecraft
Recraft V4
A proprietary design model family with raster, vector, and 4MP Pro variants.
Read model guideBlack Forest Labs
FLUX.2 Dev
An open-weight FLUX.2 model for generation plus single-reference and multi-reference editing.
Read model guideStability AI
Stable Diffusion 3.5 Large
An 8B-parameter downloadable image model for prompt adherence and customizable workflows.
Read model guideMidjourney
Midjourney V7
A hosted Midjourney model with Draft Mode, personalization, and Omni Reference workflows.
Read model guideAdobe
Adobe Firefly Image Model 4
A hosted Adobe image model for rapid ideation, realism, and creative controls.
Read model guideGoogle DeepMind
Imagen 4
A hosted text-to-image model for detailed, photorealistic, and artistic output up to 2K.
Read model guideIdeogram
Ideogram 4.0
A 9.3B open-weight text-to-image model with multilingual text and bounding-box layout control.
Read model guideByteDance Seed
Seedream 4.0
A unified hosted model for generation, editing, multiple references, and output up to 4K.
Read model guideRecraft
Recraft V4.1
A proprietary Recraft family with expressive, Utility, vector, and 4MP Pro variants.
Read model guideVERIFIED OUTPUT EXAMPLES
One local image. One local enhancement.
These repository test outputs show the current Creative Lite and Enhance Lite flow. Results vary by prompt and device.
