Krea 2 Large

Max size

Input → output

text, image → image

Weights

–

Providers

1 provider · price, latency and uptime

ProviderPer imageLatency p50Uptime · hourly, 24hRetention

Krea AI

direct API

$0.06–$0.0726 s

n<100

may retain

Listed price: $0.06–$0.07 / image
Most per image: $0.07
Precision: –
Regions: default
Headquarters: –
Data retention: May retain prompts

Supported parameters

npromptseedstylesimage_style_referencesaspect_ratioresolutioncreativityintensitycomplexitymovementmoodboardsimage_urlstrength

Uptime · last 3 days10-13 11:00 – 10-16 10:07 UTC

TueWedThuNow

By default the router tries our listings in routing order and fails over before the first byte — never mid-answer. Uptime is our own measurement; an hour stays grey until it has a sample. Open the row for the provider's specs and policies.

Quickstart

any SDK — only the base URL changes

curl --max-time 480 https://api.routerplus.com/v1/images/generations \
  -H "Authorization: Bearer $TM_API_KEY" -H "content-type: application/json" \
  -d '{"model":"krea/krea-2-large","prompt":"a red dot","aspect_ratio":"1:1","resolution":"1K"}'

# OpenAI SDK — only the base URL changes
import base64, os
from openai import OpenAI
client = OpenAI(base_url="https://api.routerplus.com/v1", api_key=os.environ["TM_API_KEY"], timeout=480.0, max_retries=0)
r = client.images.generate(model="krea/krea-2-large", prompt="a red dot",
  extra_body={"aspect_ratio": "1:1", "resolution": "1K"})
image_bytes = base64.b64decode(r.data[0].b64_json)

import OpenAI from "openai";

const client = new OpenAI({ baseURL: "https://api.routerplus.com/v1", apiKey: process.env.TM_API_KEY, timeout: 480000, maxRetries: 0 });
const r = await client.images.generate({ model: "krea/krea-2-large", prompt: "a red dot", aspect_ratio: "1:1", resolution: "1K" });
const imageBytes = Buffer.from(r.data[0].b64_json, "base64");

.qs pre[hidden] { display: block; border-top: 1px solid var(--code-line); } .qsb { display: none; } .qcode::before { content: attr(data-lang); display: block; color: var(--code-muted); font-size: 11px; margin-bottom: 6px; }

Pricing

billed at the cost Krea AI reports · at most $0.07 per image

output image	$0.06 / image · k2BillingTier=text-to-image
output image	$0.065 / image · k2BillingTier=srefs
output image	$0.07 / image · k2BillingTier=moodboards

Every billed response carries usage.cost — recompute your bill from the wire.

Parameters

n — integer 1–1 (default 1)

prompt (required) — string

seed — number or null

styles — array · item fields: id, strength · Styles (typically LoRAs) to use for the generation

image_style_references — array · up to 10 items · item fields: strength, url · Image style references. (0 = no style influence, 1 = maximum). Default 0.5.

aspect_ratio (required) — 1:1 · 4:3 · 3:2 · 16:9 · 2.35:1 · 4:5 · 3:4 · 2:3 · 9:16 (default 1:1) · default "1:1"

resolution (required) — 1K (default 1K) · default "1K"

creativity — raw · low · medium · high (default low) · default "low"

intensity — integer · -100–100 · K2 Intensity slider value from -100 to 100. Use 0 for neutral/no slider LoRA. · default 0

complexity — integer · -100–100 · K2 Complexity slider value from -100 to 100. Use 0 for neutral/no slider LoRA. · default 0

movement — integer · -100–100 · K2 Movement slider value from -100 to 100. Use 0 for neutral/no slider LoRA. · default 0

moodboards — array · up to 1 items · item fields: id, strength · Moodboards to use for generation. Currently limited to one moodboard.

image_url — string or null · Optional source image for image-to-image generation. When provided, generation starts from this image instead of pure noise. Control how much it changes with `strength`.

strength — number · 0–1 · Denoising strength when `image_url` is provided. 0 keeps the input image, 1 fully replaces it (the input image is ignored). Has no effect without `image_url`. Default 0.99. · default 0.99

Values outside these are refused with a 400 that names the field.

Uptime

10-13 11:00 – 10-16 10:07 UTC

Last 3 days–last 24h · TTFB p50 26,385 ms · uptime n<100

TueWedThuNow

An hour counts at the best provider's uptime: a request goes to another provider when one fails. Grey hours have no sample yet.