Skip to content

The split that matters most in image generation: cloud tools you use in a browser, versus open-weight models you can download and run on your own graphics card. Open-weight describes the license, not the only way to use a model, because most open models are also offered as paid cloud APIs.

As of the snapshot date, the leading closed models are GPT Image 2 (inside ChatGPT), Midjourney V8, and Google Nano Banana Pro. The leading open-weight models are FLUX.2, Qwen-Image, Ideogram 4, and Z-Image, all runnable on consumer or prosumer GPUs.

Free image generation is easy to find. Ideogram gives 10 credits a day, Recraft 30, Leonardo 150 tokens, and ChatGPT and Gemini both generate images on their free tiers. Midjourney is the notable exception, with no free tier at all.

Snapshot of 7 June 2026. 22 entries in this group, about 19 with a free option and 7 that can run on your own hardware. Prices and versions change often, and each name links to its official page. The full directory has search and filters across all categories.

How to Choose

Just make images now:

ChatGPT (GPT Image 2) and Gemini generate on their free tiers with nothing to install.

Text inside the image:

Ideogram, Qwen-Image, and Recraft specialize in clean lettering and typography. Recraft also outputs vector SVG.

Commercial safety:

Adobe Firefly is trained on licensed and Adobe Stock data and is positioned for commercial use, with Creative Cloud integration.

Run it on your own GPU:

FLUX.2 klein runs on a gaming GPU with about 13 GB of VRAM, Z-Image on about 16 GB, and Stable Diffusion 3.5 from about 8 GB. No per-image fee after the hardware.

Highest aesthetic polish:

Midjourney V8, subscription only, from 10 dollars a month.

All 22 Image generation Entries

Tool Maker What it is Cost Where it runs Hardware
Adobe Firefly Adobe Trained on licensed data for commercial safety. Creative Cloud integration. Free + paid Cloud Any device
Canva (Magic Media) Canva Image generation built into the Canva design editor. Free + paid Cloud Any device
FLUX.2 [dev] Black Forest Labs Open 32B model for generation and single or multi-reference editing. Open-weight Cloud or local High-end GPU (RTX 4090/5090)
FLUX.2 [klein] Black Forest Labs Fastest FLUX, sub-second generation on consumer hardware. Open-weight Cloud or local Gaming GPU ~13GB
FLUX.2 [pro] Black Forest Labs Top-tier photorealism and prompt adherence. The hosted commercial tier. Paid Cloud Any device
GPT Image 2 OpenAI Reasoning-based generation with near-perfect text rendering and 4K output. Free + paid Cloud Any device
Grok Imagine xAI Image and video generation inside Grok and X. Free + paid Cloud Any device
HunyuanImage 3.0 Tencent Largest open text-to-image model. Not runnable on a single consumer GPU. Open-weight Local Data-center GPUs (~3x80GB)
Ideogram 4 Ideogram Best-in-class in-image text and typography. Now open-weight. Open-weight Cloud or local High-end GPU
Imagen 4 Google Strong photorealism and text rendering, via the Gemini API. Free + paid Cloud Any device
Krea Krea AI Aggregator of 150+ models plus a real-time canvas. Free + paid Cloud Any device
Leonardo.Ai Leonardo (Canva) Game and asset-oriented platform with its own models. Free + paid Cloud Any device
Magnific (Freepik) Freepik Aggregator running its own Mystic model plus 30+ third-party models. Free + paid Cloud Any device
Midjourney Midjourney Strongest aesthetic quality and atmosphere. Discord and web app. Paid Cloud Any device
Nano Banana (Gemini Image) Google Conversational image generation and editing in Gemini. Pro adds high-res. Free + paid Cloud Any device
Playground Playground Design-focused editor routing to several top models. Free + paid Cloud Any device
Qwen-Image Alibaba Open model with excellent complex text rendering, including Chinese. Open-weight Cloud or local High-end GPU (20B)
Recraft Recraft Raster plus native vector and SVG output, with brand style sets. Free + paid Cloud Any device
Reve Reve AI Strong prompt adherence and typography from a Palo Alto startup. Free + paid Cloud Any device
Seedream ByteDance High-resolution (up to 4K) generation and editing. Paid Cloud Any device
Stable Diffusion 3.5 Stability AI The largest open LoRA and fine-tune ecosystem. Runs locally. Open-weight Cloud or local Gaming GPU 8GB+
Z-Image Alibaba Efficient 6B open model with top open photorealism. Runs on consumer cards. Open-weight Cloud or local Gaming GPU ~16GB

Common Questions

Can I generate AI images for free?

Yes. ChatGPT and Gemini generate on free tiers, Ideogram gives 10 credits a day, Recraft 30 a day (public, non-commercial), and Leonardo 150 tokens a day. Free tiers usually publish your images publicly or restrict commercial use, so read the plan terms.

What do I need to run image models locally?

A gaming GPU. Stable Diffusion 3.5 runs from about 8 GB of VRAM, FLUX.2 klein needs about 13 GB, and Z-Image about 16 GB. Once running, there is no per-image cost.

Can I use the images commercially?

It depends on the tool and the plan. Several free tiers exclude commercial use (Recraft's free plan, for example), while Adobe Firefly is explicitly positioned as commercially safe. Check the license of the specific tool and tier before publishing.