Kandinsky-5.0-T2I-Lite-Pretrainvs.Lumina-Image 2.0

Which one is better?

In our own test both are on par. Neither is in the arena. Neither has an API price.

Test images

Same task, the first image of each. The task with the biggest difference is preselected.

Kandinsky-5.0-T2I-Lite-Pretrain: Product photography, first image from the fixed test run — Kandinsky-5.0-T2I-Lite-Pretrain58 out of 100

Lumina-Image 2.0: Product photography, first image from the fixed test run — Lumina-Image 2.016 out of 100

Result for “Product photography”:Kandinsky-5.0-T2I-Lite-Pretrain ahead

Prompt given to both modelsCreate a square premium product photograph of a brushed titanium wristwatch standing upright on dark green marble. The watch has a cream dial, thin black hands set to 10:10, twelve distinct hour markers, a realistic crown, and a dark brown leather strap. Soft window light from the left, controlled reflections, shallow depth of field, no text, no logo, no extra objects.

What the rating looks at: Materials, reflections, and fine details.

“Ahead”: better in all three rating rounds. “Slightly ahead”: a higher value, but not in every round.

Test protocol of the images

Each test attempt uses the same archived prompt and one request. The first output of that attempt counts. Repeats after technical test failures are disclosed for the affected images. Files are normalized to 1,024 × 1,024 pixels without cropping. Provider resolution, randomness, quality tier and prompt processing can still differ. The images are a practical visual test, not a reconstruction of the arena configuration.

Character and detail

Kandinsky-5.0-T2I-Lite-Pretrain: manual, local PyTorch 2.13.0, Diffusers 0.36.0; kandinskylab/Kandinsky-5.0-T2I-Lite-pretrain-Diffusers@0df26d22e3290e9778b0471457d651a51d83ae3d; native 1024x1024, 1:1; one request, first output; steps=50, guidance=3.5, seed=893590444446603478; mps for text encoder and denoising; cpu for native VAE decode; bfloat16 for text encoders; float32 for transformer and CPU VAE; native tokenizer truncation: CLIP 80 to 77 tokens; archived input unchanged, generated Nov 2, 2026, 8:56 PM. Source

Lumina-Image 2.0: fal, requested 1024x1024, n=1, seed=provider random, estimated cost $0.150000, generated Nov 1, 2026, 9:38 PM. Source

Protocol gradually-image-comparison-v1. Prompt SHA 256: 985a75b38642a9f0e5e0c0e0c67f1b7f67780d42bd2168fa24e9a7a7a94b0da6

Kandinsky-5.0-T2I-Lite-Pretrain image SHA 256: af9bdf8492b30fdaeb79b8f51d168a948d43ee1bbb29107f50a2f794db91a2e0

Lumina-Image 2.0 image SHA 256: 8747df0401d91598486735bfff5d6bac6c139727a8bc50c60bf7f697f678b85f

Infographic

Kandinsky-5.0-T2I-Lite-Pretrain: manual, local PyTorch 2.13.0, Diffusers 0.36.0; kandinskylab/Kandinsky-5.0-T2I-Lite-pretrain-Diffusers@0df26d22e3290e9778b0471457d651a51d83ae3d; native 1024x1024, 1:1; one request, first output; steps=50, guidance=3.5, seed=6008496123190594626; mps for text encoder and denoising; cpu for native VAE decode; bfloat16 for text encoders; float32 for transformer and CPU VAE, generated Nov 2, 2026, 9:08 PM. Source

Lumina-Image 2.0: fal, requested 1024x1024, n=1, seed=provider random, estimated cost $0.150000, generated Nov 1, 2026, 9:38 PM. Source

Protocol gradually-image-comparison-v1. Prompt SHA 256: ed65cb4e6a51e2336f8b149ad4cbb16e020f07d527ad94362e719d2dcf91d94f

Kandinsky-5.0-T2I-Lite-Pretrain image SHA 256: 6d1a99203c55f6a9ba5853aae207dfc5e264adc73b336253fae874f0243bd60e

Lumina-Image 2.0 image SHA 256: b937b28f3c3209280c5828e18c85341d3ca443175020ca1e407dae035b5df14f

Product photography

Kandinsky-5.0-T2I-Lite-Pretrain: manual, local PyTorch 2.13.0, Diffusers 0.36.0; kandinskylab/Kandinsky-5.0-T2I-Lite-pretrain-Diffusers@0df26d22e3290e9778b0471457d651a51d83ae3d; native 1024x1024, 1:1; one request, first output; steps=50, guidance=3.5, seed=8854898730028115945; mps for text encoder and denoising; cpu for native VAE decode; bfloat16 for text encoders; float32 for transformer and CPU VAE; native tokenizer truncation: CLIP 80 to 77 tokens; archived input unchanged, generated Nov 2, 2026, 8:46 PM. Source

Lumina-Image 2.0: fal, requested 1024x1024, n=1, seed=provider random, estimated cost $0.150000, generated Nov 1, 2026, 9:38 PM. Source

Protocol gradually-image-comparison-v1. Prompt SHA 256: c67fd71e91304458a2feef5c91efa8677260fab0c0099a29a2c52cb579f5b2f3

Kandinsky-5.0-T2I-Lite-Pretrain image SHA 256: 83169f5969d6ed636c5ba8ff01236290417a5942deec1089b4725cfeb0dc2b76

Lumina-Image 2.0 image SHA 256: 74b79fcbad7565e29995c941d2b488d2edde884b307ed99e47fad994424e9dfc

Typography and layout

Kandinsky-5.0-T2I-Lite-Pretrain: manual, local PyTorch 2.13.0, Diffusers 0.36.0; kandinskylab/Kandinsky-5.0-T2I-Lite-pretrain-Diffusers@0df26d22e3290e9778b0471457d651a51d83ae3d; native 1024x1024, 1:1; one request, first output; steps=50, guidance=3.5, seed=9114026266664490789; mps for text encoder and denoising; cpu for native VAE decode; bfloat16 for text encoders; float32 for transformer and CPU VAE, generated Nov 2, 2026, 8:36 PM. Source

Lumina-Image 2.0: fal, requested 1024x1024, n=1, seed=provider random, estimated cost $0.150000, generated Nov 1, 2026, 9:37 PM. Source

Protocol gradually-image-comparison-v1. Prompt SHA 256: 2e012c3d14ee5dd02203a94f394ca80a5038cdebd6032b25a8b8a221f2255419

Kandinsky-5.0-T2I-Lite-Pretrain image SHA 256: 89cd2a1137d4d2ceb46c44b030f889bb3b14ac96ae190d2eb271dff02cfb5b74

Lumina-Image 2.0 image SHA 256: f132c159d659fb76c94cd82f1decb617c475b9c4095b177d02b73f09f2954df0

The numbers at a glance

Arena ratingfrom votes by users · higher is better

n/a

Kandinsky-5.0-T2I-Lite-Pretrain

not in the arena

n/a

Lumina-Image 2.0

not in the arena

GraduallyBenchour own rating, 0 to 100 · higher is better

Kandinsky-5.0-T2I-Lite-Pretrain

Rank 86 of 108

Lumina-Image 2.0

Rank 97 of 108

Price per imagedollars through the API · lower is better

n/a

Kandinsky-5.0-T2I-Lite-Pretrain

no API price

n/a

Lumina-Image 2.0

no API price

Ranks among the 21 image models with an arena rating and the 108 with a GraduallyBench run; prices compared with 28 models with an API price.

Kandinsky-5.0-T2I-Lite-Pretrain

Subscription: n/a No plan in Germany names exactly this model.
Via API: No API price known. Access: local.
License: Open weights.

Lumina-Image 2.0

Subscription: n/a No plan in Germany names exactly this model.
Via API: No API price known. Access: local.
License: Open weights.

Subscriptions only count plans that name exactly this model. Subscription prices include VAT, API prices are in dollars.

GraduallyBench from Nov 3, 2026 · Subscription database from Sep 29, 2026 (38 days old)

All data side by side

The better value is bold. For ratings only when the gap is clear.

Kennzahl	Kandinsky-5.0-T2I-Lite-Pretrain	Lumina-Image 2.0
Ratings
Arena ratingfrom votes by users, higher is better	n/a	n/a
Arena rank	n/a	n/a
Arena votesmore votes, a safer rating	n/a	n/a
GraduallyBenchour own rating, 0 to 100	28	18
GraduallyBench by task
Typography and layoutLegible text, hierarchy, and composition	16	18
Product photographyMaterials, reflections, and fine details	58	16
Character and detailAnatomy, clothing, and spatial consistency	14	23
InfographicNumbers, labels, and visual organization	23	16
API price
Per imagelower is better	n/a	n/a
100 images	n/a	n/a
Profile
Released	n/a	Jan 25, 2025
Weightsopen means: runs locally	open	open
Access	local	local
Parameters	6 billion	2.6 billion
Largest output	not consistently published	not consistently published
Plans in Germanynaming exactly this model	n/a	n/a

API prices for one image at 1,024 × 1,024 pixels according to Artificial Analysis.

Sources: Kandinsky-5.0-T2I-Lite-Pretrain model source (verified Nov 1, 2026) · Lumina-Image 2.0 model source (verified Nov 1, 2026)

Arena ratings by category

From blind votes on LMArena. “Ahead” means the lead is clear.

No shared arena score in the stored snapshot

The stored benchmark database contains no measurement for Kandinsky-5.0-T2I-Lite-Pretrain and Lumina-Image 2.0 from the same arena and category. Older model versions are not used as substitutes.

More comparisons

Compare Kandinsky-5.0-T2I-Lite-Pretrain and Lumina-Image 2.0 with the models people compare them with most often.

Compare another matchup

First image modelSecond image model