Krea V2 Turbo LoRA API Documentation

Krea V2 Turbo LoRA API Documentation

Playground

Krea 2 Turbo with LoRA support runs custom LoRAs (trained on Krea 2 Raw, comfy or diffusers naming) on the 8-step distilled Krea 2 engine, generating personalized images natively at 1K to 2K resolution, with both text-to-image and image-to-image supported. Ready-to-use REST inference API, best performance, no cold starts, affordable pricing.

Features

Krea 2 Turbo is Krea’s first from-scratch foundation image model, released as open weights. The Turbo checkpoint is distilled to 8 sampling steps and generates images natively from 1K up to 2K resolution with the aesthetic-first look Krea is known for — no oversaturated, over-smooth “AI look”. This endpoint adds LoRA loading: bring up to 3 LoRA URLs (trained on Krea 2 Raw; ComfyUI and diffusers/PEFT naming both work) and they are applied on the fly.

Why Choose This?

• Native 2K generation — 2048x2048 output straight from the model, no upscaler pass.
• 8-step Turbo speed — distilled inference engine; images in seconds.
• Aesthetic-first — trained for photographic realism and strong art direction.
• Text-to-image and image-to-image — pass an image URL to restyle or vary it, with strength control.
• Up to 3 LoRAs — each with its own scale; diffusers-format LoRAs are converted automatically.
• Flexible aspect ratios — 15 presets from 1:3 to 3:1, plus vertical and widescreen formats.

Parameters

Pricing

Pro Tips

• Krea 2 responds well to photographic language: film stock, lens, lighting terms.
• For image-to-image, strength 0.4-0.6 keeps composition; 0.7-0.9 repaints more freely.
• 2K is generated natively by the model — use it for final assets; 1K for iteration.
• LoRA scale around 0.8-1.0 is a good starting point; lower it if the style overpowers the prompt.

Related Models

• Krea 2 Turbo — the base engine without LoRA loading.

Authentication

For authentication details, please refer to the Authentication Guide.

API Endpoints

Submit Task & Query Result

set -euo pipefail

export WAVESPEED_API_KEY="your-api-key"

REQUEST_BODY=$(cat <<'JSON'
{
  "prompt": "A cinematic ocean wave at sunrise, highly detailed",
  "aspect_ratio": "1:1",
  "resolution": "1k",
  "strength": 0.8,
  "seed": -1
}
JSON
)

# 1. Submit the prediction.
SUBMIT_RESPONSE=$(curl --silent --show-error --fail-with-body \
  -X POST "https://api.wavespeed.ai/api/v3/wavespeed-ai/krea-v2/turbo-lora" \
  -H "Authorization: Bearer ${WAVESPEED_API_KEY}" \
  -H "Content-Type: application/json" \
  -d "${REQUEST_BODY}")

TASK=$(printf '%s' "${SUBMIT_RESPONSE}" | jq 'if type == "object" and has("data") then .data else . end')
PREDICTION_ID=$(printf '%s' "${TASK}" | jq -r '.id // empty')
if [ -z "${PREDICTION_ID}" ]; then
  printf 'Submission response did not contain a prediction id
' >&2
  exit 1
fi
RESULT_URL=$(printf '%s' "${TASK}" | jq -r '.urls.get // empty')
if [ -z "${RESULT_URL}" ]; then RESULT_URL="https://api.wavespeed.ai/api/v3/predictions/${PREDICTION_ID}/result"; fi

# 2. Poll until the prediction finishes.
while true; do
  RESPONSE=$(curl --silent --show-error --fail-with-body \
    "${RESULT_URL}" \
    -H "Authorization: Bearer ${WAVESPEED_API_KEY}")
  RESULT=$(printf '%s' "${RESPONSE}" | jq 'if type == "object" and has("data") then .data else . end')
  STATUS=$(printf '%s' "${RESULT}" | jq -r '.status // empty')

  case "${STATUS}" in
    completed) printf '%s\n' "${RESULT}" | jq '.outputs'; break ;;
    failed|cancelled|timeout) printf '%s\n' "${RESULT}" | jq . >&2; exit 1 ;;
    created|processing) sleep 2 ;;
    *) printf 'Unexpected status: %s
' "${STATUS}" >&2; exit 1 ;;
  esac
done

Parameters

Task Submission Parameters

Request Parameters

Response Parameters

Result Request Parameters

Result Response Parameters