catalog

Models

Every model on the Supernet, priced per render. Filter by what you need, click through for specs, live quotes, and working code.

25 curated models + 374 community checkpoints

Sample output from Z-Image Turbo
image

Z-Image Turbo

Tongyi Lab

The default. Sub-cent renders in seconds — and the only model free Spark can drive via API.

from 0.80 ($0.0040) · 1024² · 4 steps
Sample output from Z-Image
image

Z-Image

Tongyi Lab

The full-step sibling of Turbo — same model line, maximum detail.

from 4.00 ($0.02) · 1024² · 25 steps
Sample output from FLUX.2 [dev]
image

FLUX.2 [dev]

Black Forest Labs

Black Forest Labs’ flagship — top-tier fidelity with up to six reference images.

from 4.15 ($0.021) · 1024² · 20 steps
Sample output from FLUX.1 [schnell]
image

FLUX.1 [schnell]

Black Forest Labs

The speed classic — four-step FLUX for instant iteration.

from 0.80 ($0.0040) · 1024² · 4 steps
Sample output from FLUX.1 Krea [dev]
image

FLUX.1 Krea [dev]

premium spark

Black Forest Labs × Krea

Aesthetic-tuned FLUX — opinionated, photographic, anti-generic.

from 4.53 ($0.023) · 1024² · 20 steps
F
image

FLUX.1 Kontext [dev]

premium spark

Black Forest Labs

Reference-driven editing — change one thing, keep everything else.

from 4.15 ($0.021) · 1024² · 20 steps
Sample output from Chroma v.46 [flash]
image

Chroma v.46 [flash]

Lodestone Rock

Flux-family community flagship at half the flagship price.

from 2.20 ($0.011) · 1024² · 10 steps
Sample output from Chroma v.48 [detail]
image

Chroma v.48 [detail]

Lodestone Rock

The detail-calibrated Chroma — texture-rich final renders.

from 4.18 ($0.021) · 1024² · 20 steps
Sample output from Qwen Image 2512
image

Qwen Image 2512

Alibaba (Qwen)

Qwen’s image flagship — exceptional at rendering legible text.

from 4.15 ($0.021) · 1024² · 20 steps
Sample output from Qwen Image 2512 Lightning
image

Qwen Image 2512 Lightning

Alibaba (Qwen)

Four-step Qwen — flagship text rendering at draft speed.

from 2.07 ($0.01) · 1024² · 4 steps
Sample output from Qwen Image Edit 2511
image

Qwen Image Edit 2511

Alibaba (Qwen)

The strongest preservation editor — same face, same scene, new instruction.

from 4.15 ($0.021) · 1024² · 20 steps
Sample output from Qwen Image Edit 2511 Lightning
image

Qwen Image Edit 2511 Lightning

Alibaba (Qwen)

The default editor in Sogni’s creative tools — fast, faithful edits.

from 2.10 ($0.011) · 1024² · 4 steps
G
imagepartner

GPT Image 2

OpenAI

OpenAI’s image model through the Sogni API — 16 references, superb instructions.

from 13.7 ($0.069) · 1024² · medium
video

LTX-2.3 22B Distilled

Lightricks

The default video model — fast, cheap, and it speaks. Native audio at $0.07 per 5s clip.

from 13.8 ($0.069) · 5s · 720p
video

LTX-2.3 22B Dev

Lightricks

Full-step LTX for longer, steadier shots — the quality tier of the default.

from 34.6 ($0.173) · 5s · 720p
video

WAN 2.2 14B LightX2V

Alibaba (Wan)

The character specialist — lip-sync, motion transfer, and subject replacement.

from 46.2 ($0.231) · 5s · 720p
video

WAN 2.2 14B

Alibaba (Wan)

Full-step WAN — maximum quality from the WAN line.

from 115.4 ($0.577) · 5s · 720p
videopartner

Seedance 2.0

ByteDance

ByteDance’s frontier video model — the first partner model on the Sogni API.

from 197.7 ($0.988) · 5s · 720p
videopartner

Seedance 2.0 Fast

ByteDance

Seedance speed tier — frontier motion, 720p cap, ~20% cheaper.

from 156.1 ($0.781) · 5s · 720p
A
music

ACE-Step 1.5 Turbo

ACE Studio × StepFun

Full songs with vocals for half a cent per 30 seconds.

from 0.90 ($0.0043) · 30s · 8 steps
A
music

ACE-Step 1.5 SFT

ACE Studio × StepFun

The quality tier — tighter lyric adherence, more sampler control.

from 4.87 ($0.024) · 30s · 50 steps
A
music

ACE-Step 1.5 XL Turbo

ACE Studio × StepFun

The larger XL model on the turbo schedule — fuller songs, still cents per track.

from 1.08 ($0.0054) · 30s · 8 steps
A
music

ACE-Step 1.5 XL SFT

ACE Studio × StepFun

The XL model at full fifty-step quality — the top music tier in the catalog.

from 6.08 ($0.03) · 30s · 50 steps
Q
language

Qwen3.6 35B-A3B

Alibaba (Qwen)

The default LLM — reasoning, tools, and vision behind an OpenAI-compatible endpoint.

from 60 ($0.30) · per 1M tokens
Q
language

Qwen3.5 35B-A3B Abliterated

Alibaba (Qwen) · community abliteration

The uncensored variant — for applications that need an unfiltered model.

from 60 ($0.30) · per 1M tokens

collection

SDXL & SD Community

374 community checkpoints — Juggernaut, DreamShaper, CyberRealistic, Animagine, Pony and more. From 0.11 ◈ on the Relaxed network.

Browse the collection →