NightCafe AI Models Guide

NightCafe AI Models Guide: Choosing the Right Model

Overview: How to choose a NightCafe model

NightCafe offers dozens of AI models for creating images and videos. Each model has different strengths — some excel at photorealism, others at typography, artistic styles, editing existing images, or generating video. This guide lists NightCafe's official models, highlights the best model for common tasks, and explains what the newest models are best at.

A few things to know before you pick a model:

  • Newest & recommended models appear first in the model picker (and near the top of this guide). These are the models we suggest for most creators.
  • Approx. credits shown below is the base credit cost for one standard image. Final cost varies with resolution, number of images, start/reference images, and any PRO unlimited allowances.
  • PRO means the model requires a NightCafe PRO subscription. Some models are free or included with lower credit costs.
  • Editing capabilities: "Prompt-to-edit" means you can edit an existing image with a text instruction. "Multi-image fusion" means you can combine several reference images into one result. "Consistent character" means the model can keep a character or subject consistent across multiple images (great for comics and storyboards).

Best model for each task (quick picks)

Use this section as a fast lookup. All picks below are current, available models. If a recommended model is PRO-only, a lower-cost alternative is noted where possible.

Task / use-caseBest picksWhy
All-round best qualityGPT Image 2 (Medium/High), Nano Banana Pro, Nano Banana 2, Seedream 4.5Top scores across prompt accuracy, realism and text. Great default choices.
Best value / low costZ-Image Turbo (~0.5), Flux Schnell (~0.5), GPT Image 2 Low (~0.6), GPT Image 1.5 Low (~1)Strong quality for very few credits — ideal for drafts and high-volume work.
FreeStable Diffusion 1.5, DreamShaper v8Free base generations, no PRO required. Older SD1.5 quality.
PhotorealismGPT Image 2 (High/Medium), Nano Banana 2 / Pro, Seedream 4.5, Grok Imagine Image Quality, Flux 2 Pro, Qwen Image 2512Lifelike people, skin, lighting and natural detail.
Text & typographyGPT Image 2, Nano Banana Pro, Nano Banana 2, Recraft v3, Ideogram V3, Qwen Image / 2.0, Flux 2 FlexAccurate spelling and layout for posters, signs and captions.
Logos, graphic design & vectorsRecraft v3 (can output vectors), Ideogram V3, Flux 2 FlexClean design output; Recraft v3 can generate scalable vectors.
Artistic / creative stylesSeedream 4.x, GPT Image 2 (Medium/High), Flux 2 Klein 9B, HiDream I1, Stable CorePainterly, stylised and concept-art looks.
Complex prompts / prompt adherenceGPT Image 2 High, Nano Banana Pro, GPT Image 2 Medium, Seedream 4.x, Grok Imagine Image QualityFollows long, detailed instructions faithfully.
Editing an existing image (prompt-to-edit)Nano Banana 2 / Pro, GPT Image 2, Flux Kontext (Pro/Max), Qwen Image Edit, Seedream 4.xChange or retouch a photo using a text instruction.
Combining reference images (multi-image fusion)GPT Image 2, Nano Banana 2 / Pro, Seedream 4.x, Flux 2 family, Qwen Image 2.0Merge several inputs (e.g. product + scene) into one image.
Consistent characters (comics, storyboards)Nano Banana Pro / 2, GPT Image 2, Seedream 4.x, Flux 2 familyKeep the same character across multiple images.
Anime & illustrationAnimagine XL, Blue Pencil XL, AAM XL Anime Mix (SDXL); Seedream for modern stylised looksDedicated anime checkpoints plus flexible modern models.
Highest resolution (4K)Seedream 4.0 / 4.5 (up to 4K), Nano Banana 2 (up to 4K), Flux PRO 1.1 Ultra (2K)Large, print-ready output.
Grounded generation / reasoningGPT Image 2, Nano Banana Pro / 2, Flux 2 Max, Gemini Flash 2.5Uses real-world context and reasoning for accurate results.
VideoGoogle Veo 3.1 (with audio), Seedance 2.0 / 1.5 Pro, Kling 3.0 Turbo, Grok Imagine Video, PixVerse V6Text-to-video and image-to-video, several with synchronised audio.
Custom subject / your own styleFine-tuned LoRA models (see the LoRA section)Train a model on your own face, object or style.

Latest & recommended image models (detailed)

These are the newest, highest-quality models and the ones NightCafe recommends first. They power the best results for most creators.

OpenAI — GPT Image 2 & 1.5

OpenAI's GPT Image models are among the strongest all-rounders on NightCafe, with class-leading prompt fidelity, text rendering and layout control. They support prompt-to-edit, multi-image fusion, consistent characters and grounded reasoning. Quality tiers (Low / Medium / High) trade cost for detail.

  • GPT Image 2 Low (~0.6 credits, PRO) — excellent value with strong prompt fidelity and near-perfect text. A recommended everyday model.
  • GPT Image 2 Medium (~5.3 credits, PRO) — flagship quality: outstanding prompt accuracy, realism and typography.
  • GPT Image 2 High (~21 credits, PRO) — the highest-fidelity setting for exceptional detail and text.
  • GPT Image 1.5 (Low / Medium / High) (~1 / 3.5 / 10 credits, PRO) — the previous OpenAI generation, still capable and supports editing and fusion.
  • GPT Image 1 (Low / Medium / High) — earlier OpenAI generation, superseded by 1.5 and 2.

Google — Nano Banana (Gemini)

Google's Gemini image models (nicknamed "Nano Banana") have first-class start-image support and are excellent for editing and combining photos, with strong text rendering.

  • Nano Banana Pro (~11 credits, PRO) — the most capable Nano Banana, with perfect prompt adherence and typography scores, plus reasoning and consistent characters.
  • Nano Banana 2 (~7 credits, PRO) — AKA Gemini Flash 3.1; Google's latest with excellent all-round quality, 4K support, editing, fusion and reasoning.
  • Nano Banana 2 Lite (~3.5 credits, PRO) — a faster, cheaper variant for 1K text-to-image and editing.
  • Nano Banana (Gemini Flash 2.5) (~4 credits, PRO) — strong editing and fusion at a lower cost.

ByteDance — Seedream

Seedream models are fast, high-resolution and great for both realism and artistic work, with editing and fusion support.

  • Seedream 4.5 (~4.5 credits, PRO) — improved 4.0 with top realism and prompt accuracy; up to 4K.
  • Seedream 4.0 (~3.5 credits, PRO) — ultra-fast high-res (2K–4K) multimodal generation and editing.
  • Seedream 5.0 Lite (Preview) (~3.5 credits, PRO) — responsive text-to-image with precise prompt adherence; up to 3K.

Black Forest Labs — Flux

The Flux family spans fast low-cost models to high-end production tools, and includes dedicated image-editing models (Kontext). Flux 2 adds editing, fusion and consistent-character support.

  • Flux 2 Max (~7 credits, PRO) — grounded generation with real-world context; maximum quality for complex workflows.
  • Flux 2 Pro (~3 credits, PRO) — high image quality, strong photorealism.
  • Flux 2 Flex (~6 credits, PRO) — tuned for typography and designs.
  • Flux 2 Dev (~1.5 credits, PRO) — a well-rounded baseline Flux 2 at reasonable cost.
  • Flux 2 Klein 9B / 9B Fast / 4B / 4B Fast (~0.5–1 credits) — distilled models built for sub-second, near real-time generation and editing. "Klein 9B Fast" is a recommended low-cost pick.
  • Flux Kontext (Dev / Pro / Max) (~2.5 / 3.5 / 7 credits) — specialised for editing images with natural-language prompts.
  • Flux PRO 1.1 (~4 credits, PRO) — the full-power Flux 1.1; Flux PRO 1.1 Ultra (~7 credits) adds 2K output.
  • Flux Schnell (~0.5 credits) — a cheaper, faster Flux; a recommended budget model.
  • Flux Krea (~1.5 credits, PRO) — realistic images with a focus on correct human anatomy.
  • Flux (Dev) (~1 credit, PRO) — a great low-cost general-purpose model.
  • Juggernaut Flux (Base / Pro / Lightning) — Flux fine-tunes by RunDiffusion focused on realism.

Alibaba — Qwen Image & Z-Image

  • Z-Image Turbo (~0.5 credits) — a recommended, very affordable model with excellent realism for the price.
  • Qwen Image 2512 (~1 credit, PRO) — major realism update: natural humans, sharp textures, accurate text.
  • Qwen Image 2.0 / 2.0 Pro (~3.5 / 7.5 credits, PRO) — unified generation and editing with native 2K and excellent typography.
  • Qwen Image Edit / Edit Plus / Edit 2511 (~1–2 credits) — dedicated editing models that preserve people and textures while changing scenes.
  • Qwen Image / Qwen Image SD (~1 / 0.75 credits) — typography-strong base models; SD is a cheaper, lower-res option.

Ideogram & Recraft

  • Ideogram V3 / V3 Quality / V3 Turbo (~4 / 6 / 2 credits, PRO) — strong prompt adherence and text-in-image; Quality is the best, Turbo the cheapest.
  • Ideogram 2a / 2a Turbo / 2.0 / 2.0 Turbo — earlier Ideogram generations, good at typography.
  • Recraft v3 (~5 credits, PRO) — excellent spelling and typography; can generate vector graphics.

xAI — Grok Imagine

  • Grok Imagine Image Quality (~7 credits, PRO) — xAI's most capable image model: photorealistic and stylised output, native 2K, iterative refinement.
  • Grok Imagine Image (~2 credits, PRO) — high-quality images from text or a start image, with prompt-to-edit.

HiDream I1

  • HiDream I1 Full (~2 credits, PRO), I1 Dev (~1.5 credits, PRO), I1 Fast (~1 credit) — artistic, high-clarity image generation at different speed/cost points.

Stable Core

Stable Core (~4 credits, PRO) — beautiful, high-resolution images without complex prompts; a dynamic, easy-to-use option.

Video models

Video models generate short clips from a text prompt and/or a start image. Many now include synchronised audio. Video models are PRO, and most require a higher PRO plan level. Credit costs are per clip and are higher than image models.

  • Google Veo 3.1 / 3.1 Fast / 3.1 Lite — lifelike motion with context-aware synchronised audio; supports first- and last-frame references. Veo 3.1 is a popular pick.
  • Seedance 2.0 / 2.0 Fast / 2.0 Mini and Seedance 1.5 Pro — ByteDance's state-of-the-art video with native synchronised audio; 1.5 Pro is popular.
  • Kling 3.0 Turbo / 2.5 Turbo Pro / 2.5 Turbo Standard / 2.1 Standard / 1.6 Standard — cinematic motion, strong prompt adherence and improved lip sync.
  • Grok Imagine Video — xAI's multimodal video with native audio from text or a start image.
  • PixVerse V6 / V5 — multi-shot storytelling with synchronised audio and cinematic controls.
  • Happy Horse 1.0 — best-in-class motion quality with synchronised audio (Alibaba).
  • Gemini Omni Flash — Google's multimodal video model with native audio; 720p from text or a start image.
  • Runway Gen-4 Turbo — Runway's latest image-to-video model.
  • Seedance 1.0 Pro / Pro Fast — earlier Seedance video models.

LoRAs & fine-tuned (custom) models

In addition to the official models above, you can train your own custom model on NightCafe. These are LoRAs ("Low-Rank Adaptation") — a fast way to fine-tune Stable Diffusion on a small set of images so it can recreate a specific face, object/animal, or style.

How to train a LoRA

  1. Open My Models (in the main menu on desktop, or under your profile picture on mobile) and click Fine-tune a new model.
  2. Choose a model type (face, object/animal, or style) and give it a name.
  3. Choose or create a dataset — upload at least ~20 images. More varied images generally give better results.
  4. Agree to the terms and start training. Training typically takes about 10–30 minutes, and you'll be notified when it's done.

Fine-tuning is a PRO feature, but free users get 1 free face-model tune plus 10 free generations with it. Your trained models are private to your account.

How to use a LoRA in a prompt

To use a fine-tuned model, add its token to your prompt in the format <type:name:optional weight> — for example <lora:My Face:0.8>. You can write prompts like "A photo of <lora:My Face:0.8> riding an elephant" or "A unicorn in the style of <lora:Dark Fantasy:0.5>". The weight is optional (usually between 0 and 1) and defaults to 0.8 if omitted.

Learn more in Introducing Fine-Tuning on NightCafe.

Older model classes (still available)

NightCafe keeps many earlier open-source checkpoints available. These are cheaper and can be great for specific styles, but generally have lower prompt adherence and text ability than the latest models. Rather than list every checkpoint, they're grouped into classes below.

SDXL models (Stable Diffusion XL)

SDXL launched in 2023 as a major step up from SD 1.5, producing sharper 1024px images. NightCafe hosts many community SDXL checkpoints, each tuned for a look. All are low-cost (~0.5–1 credit), and "Lightning" variants are the fastest/cheapest. Examples by strength:

  • Photorealism: RealVisXL (v3–v5), Crystal Clear XL, Cherry Picker XL, Fluently XL, Juggernaut XL / v9 / XI.
  • Art & concept: DreamShaper XL, Starlight XL, Mysterious XL, Stable Core.
  • Anime & stylised: Animagine XL, Blue Pencil XL, AAM XL Anime Mix, Virtual Utopia XL.
  • Cartoon: Real Cartoon XL.
  • Base & utility: SDXL 1.0, SDXL LCM, SDXL DPO.

SD 1.5 models (Stable Diffusion 1.5)

SD 1.5 is the popular first-generation Stable Diffusion model from 2022. It's fast and inexpensive (many are free), but lower resolution and less accurate than modern models. NightCafe hosts many SD 1.5 community checkpoints, including:

  • Realism: Realistic Vision, AbsoluteReality, Juggernaut Reborn.
  • Art & "does-it-all": DreamShaper v8 (free), NeverEnding Dream.
  • Anime, comics & cartoon: Blue Pencil, Arthemy Comics, 3D Animation Diffusion, RealCartoon Pixar, Ghost Mix.
  • Niche: RPG (character portraits), Rabbit (cute art), WildlifeX (animals), Nightmare Shaper (dark art).

Upscalers

  • Clarity Upscaler — adds detail and clarity while upscaling up to 4x.
  • Creative Upscaler (PRO) — a creative upscaler capable of 4K images.

Original algorithms

NightCafe's earliest algorithms remain available for nostalgia and unique looks: Artistic (VQGAN+CLIP, the original text-to-image method), Coherent (CLIP-guided diffusion), and Style Transfer (neural style transfer).

Retired / deprecated models

Some models have been retired, usually because an external provider discontinued them. These are no longer available for new creations:

Understanding model attributes

NightCafe rates most image models on four qualities, each scored from 0 to 1 (higher is better). These power the "best for" picks above:

  • Prompt adherence — how faithfully the model follows your prompt.
  • Art — strength at artistic and stylised output.
  • Realism — photorealistic quality.
  • Typography — ability to render readable, correct text.

Base cost is the approximate credits for one standard image. Actual credits depend on resolution, number of images, start/reference images and your plan. See the credits and subscription articles for details.

FAQ

Which model should I use if I'm not sure? Start with a recommended model such as GPT Image 2 Low or Nano Banana 2 for balanced, high-quality results, or Z-Image Turbo / Flux Schnell if you want to keep costs low.

Which models are free? Stable Diffusion 1.5 and DreamShaper v8 offer free base generations. Most modern models require credits and many require PRO.

Which models can edit my photos? Use a prompt-to-edit model such as Nano Banana 2/Pro, GPT Image 2, Flux Kontext, or Qwen Image Edit.

Which models make video? See the Video models section — Veo 3.1, Seedance, Kling, Grok Imagine Video and PixVerse are current options.

Can I make my own model? Yes — train a LoRA on your own face, object or style. See the LoRA section above.

Model availability and pricing change over time as providers release new models and retire old ones. This guide reflects NightCafe's official model lineup at the time of writing. (7th July 2026)

    • Related Articles

    • Announcing the upcoming deprecation of fine-tuned SDXL models and SDXL 0.9 & BETA

      Some features on NightCafe are powered by Stability AI, the creators of Stable Diffusion. Stability AI have announced that they're deprecating the following features: Fine-tuning for SDXL will be deprecated on October 18 Stable Diffusion XL v0.9 & ...
    • Stable Diffusion AI Image Generator

      Stable Diffusion is the new darling of the AI Art world. This algorithm is one of the latest additions to NightCafe and is even more coherent than the "Coherent" algorithm. "Stable" is short for "Stable Diffusion" - an open-source algorithm and model ...
    • UPDATE: Announcing the upcoming deprecation of Rundiffusion Juggernaut models on December 15

      Update: We've found a partner with a commercial license for these models, and will continue to support the Juggernaut models on NightCafe. Juggernaut XL versions 9 and 11 are licensed for commercial use by their creator - Rundiffusion. On the 15th of ...
    • GPT Image on NightCafe

      What is GPT Image? GPT Image is OpenAI’s image model, available on NightCafe as GPT Image 1, 1.5 and 2. Each comes in Low, Medium and High quality settings — lower settings are faster and cheaper (great for prototyping), higher settings give the best ...
    • Flux 2 on NightCafe

      What is Flux 2? Flux 2 (by Black Forest Labs) is a versatile, high-quality image model available on NightCafe in several variants, so you can trade off quality, speed and cost: Flux 2 Pro — improved image quality, great at photorealism. Flux 2 Dev — ...