ARCHITECTURELM
Index
Models
Tools
Spaces
About
Model Index — The engines underneath
Flux
Black Forest Labs
Open-weight image models with the strongest geometry retention and native LoRA training.
3 models
Google
Google DeepMind
Gemini-based image generation. Reads massing well and follows long prompts literally.
2 models
Seedream
ByteDance
Fast, low-cost image generation. Built for volume option testing rather than final imagery.
1 models
Midjourney
Midjourney Inc.
Hosted-only image generation. Unmatched atmosphere, weakest geometry control.
Kling
Kuaishou
Image-to-video generation. Turns a still render into a camera move.
All models — filter by task
Label
PUBLICATION
ArchitectureLM
ISSUE
014 — Rev C
EDITION
2026
CONTACT
hello@architecturelm.com
Flux.2 Pro
INPAINT
I2I
T2I
The hosted tier. Better material realism than Klein, but you cannot train it on your own work.
Nano Banana Pro
Reads a massing model unusually well and follows long prompts literally. Weaker on material realism.
Flux.2 Klein
Strong geometry retention and trainable on your own style. The current default for serious archviz.
Kling 2.5
I2V
Previous generation. Cheaper per second, with more drift on facade detail through camera moves.
Nano Banana
The standard tier. Faster and cheaper than Pro, with noticeably looser geometry.
Flux.1 Pro
The previous generation. Still widely used because the LoRA ecosystem around it is mature.
Seedream v4.5
Fast and cheap per generation. Good for volume option testing, less so for final imagery.
T2V
Turns a still render into a credible camera move. Watch for drift on fine facade detail.
Unmatched for atmosphere and mood boards; poor at holding your actual geometry.