AI image models, compared

MooseAI puts 25 AI image models in one studio, in your browser: Google Nano Banana, OpenAI GPT Image, ByteDance Seedream, FLUX, Grok, Qwen Image and Recraft. Here is what each one can do. Open any of them to make an image from a prompt or edit your photos, then animate it into a video or upscale it.

Open the image studio

Explore image models

View all 25 models →

Black Forest Labs

FLUX 3

Resolution: 4K

Which model should I use?

Each image uses credits, and the studio shows how many before you generate.

All models side by side

ModelMade byPhotos to editTop resolutionStyles
Nano BananaGoogleUp to 6——
Nano Banana 2GoogleUp to 144K—
Nano Banana 2.1GoogleUp to 144K—
Nano Banana 2 LiteGoogleUp to 10——
Nano Banana ProGoogleUp to 144K—
GPT Image 2OpenAIUp to 4——
GPT Image 2.5OpenAIUp to 4——
GPT Image 2.6OpenAIUp to 4——
Seedream V4ByteDanceUp to 64K—
Seedream V4.5ByteDanceUp to 64K—
Seedream V5ByteDanceUp to 143K—
Seedream 5 ProByteDanceUp to 102K—
Seedream V5 FlashByteDanceUp to 10——
Seedream 5.5ByteDanceUp to 102K—
Grok ImagexAIUp to 1——
Grok Image QualityxAIUp to 12K—
Grok Image 2xAIUp to 12K—
FLUX 3Black Forest LabsUp to 104K—
Flux LightningBlack Forest Labs———
Flux 1.1 Pro UltraBlack Forest Labs———
Qwen Image 2.1AlibabaUp to 10——
Qwen Image 2.1 ProAlibabaUp to 10——
Recraft 20BRecraft——26
Recraft V3Recraft——19
Z-Image TurboAlibaba, optimized by Pruna AI———

A dash means the studio has no setting for it, and the model decides.

The model families

Nano Banana: Nano Banana, Nano Banana 2, Nano Banana 2.1, Nano Banana 2 Lite, Nano Banana Pro

Nano Banana is the popular name of Google’s Gemini image models. It is known for editing photos with plain instructions while keeping people and products consistent, and for combining several images into one. Nano Banana Pro, built on Gemini 3 Pro, adds sharper text in images and output up to 4K; Nano Banana 2 and 2.1 bring that quality at a lower cost, and Nano Banana 2 Lite is the fastest, lowest-cost version.

GPT Image: GPT Image 2, GPT Image 2.5, GPT Image 2.6

GPT Image is OpenAI’s image model, the one behind image generation in ChatGPT. It follows long, detailed prompts closely, writes legible text in images, and edits or combines the images you give it. GPT Image 2.6, released in November 2026, makes images about 50% faster and adds animated GIFs and marked-up edits.

Seedream: Seedream V4, Seedream V4.5, Seedream V5, Seedream 5 Pro, Seedream V5 Flash, Seedream 5.5

Seedream is the image model family of ByteDance’s Seed team. Since Seedream 4.0 one model both creates and edits images, takes many reference images at once and makes high-resolution output, with good text in English and Chinese.

Grok Imagine: Grok Image, Grok Image Quality, Grok Image 2

Grok Imagine is xAI’s image and video generator, the one in Grok. Its image models make photorealistic and stylized pictures from a prompt, and can start from a photo you give them.

FLUX: FLUX 3, Flux Lightning, Flux 1.1 Pro Ultra

FLUX is the image model family of Black Forest Labs, founded by the researchers behind Stable Diffusion. FLUX.1 [schnell] is a fast, open-weights model; FLUX1.1 [pro] Ultra makes large, detailed images; FLUX 3 is the newest generation, which also edits images and keeps a look from reference images.

Qwen Image: Qwen Image 2.1, Qwen Image 2.1 Pro

Qwen Image is the image model family of Alibaba’s Qwen team. It is known for rendering text in images accurately, in English and Chinese, and for precise edits from instructions and reference images.

Recraft: Recraft 20B, Recraft V3

Recraft makes image models for designers. Recraft V3 led the Hugging Face text-to-image arena at its launch in late 2024, and is known for brand-ready styles, clean composition and text in images; Recraft 20B is a smaller, lower-cost model. Both offer many styles, from photography to illustration.

Z-Image Turbo: Z-Image Turbo

Z-Image Turbo is a small, fast photorealistic image model from Alibaba’s Tongyi lab, released with open weights. Pruna AI, which works on making AI models faster and cheaper to run, serves an optimized version that makes an image in about a second.

More than images

Every image you make can become a video with one of the AI video models, or a larger copy with the AI image upscaler.