Flux 2 Klein 9b Text To Image LoRA API Documentation

Flux 2 Klein 9b Text To Image LoRA API Documentation

Playground

FLUX.2 [klein] 9B with LoRA support is a high-quality text-to-image model with 9B parameters, offering enhanced realism, crisper text generation, and fast LoRA customization. Ready-to-use REST inference API, best performance, no cold starts, affordable pricing.

Features

FLUX.2 Klein 9B Text-to-Image LoRA is a powerful text-to-image generation model with full LoRA support. With 9B parameters, it delivers higher quality and detail compared to the 4B variant while maintaining fast generation. Apply custom LoRA adapters for personalized styles and characters.

Why Choose This?

• Enhanced quality 9B parameter model delivers richer detail and better prompt understanding than the 4B variant.

• LoRA support Apply custom LoRA adapters for personalized styles, characters, or visual aesthetics.

• Flexible sizing Custom width and height controls for any aspect ratio.

• Prompt Enhancer Built-in tool to automatically improve your prompts for better results.

• Balanced performance More capable than 4B while remaining fast and affordable.

Parameters

LoRA Format

Each LoRA in the loras array has:

• path (required) — URL to the LoRA weights file
• scale (optional) — Weight multiplier, default 1

How to Use

1. Write your prompt — describe the image including subject, style, lighting, and mood.
2. Set size — adjust width and height for your desired dimensions.
3. Add LoRAs (optional) — click ”+ Add Item” to include custom LoRA adapters.
4. Set seed — use -1 for random, or specify a number for reproducibility.
5. Run — submit and download the generated image.

Pricing

Simple flat-rate pricing regardless of image size or LoRA count.

Best Use Cases

• Custom Character Generation — Apply character LoRAs for consistent identity.
• Style Personalization — Use style LoRAs for unique visual aesthetics.
• Brand Content — Create on-brand visuals with trained LoRA styles.
• High-Quality Output — When 4B quality isn’t enough but full-size models are overkill.
• Production Work — Balanced quality and cost for professional content.

Pro Tips

• Be specific in your prompts — include subject, style, lighting, colors, and atmosphere.
• Start with LoRA scale 1.0 and adjust based on results.
• Combine multiple LoRAs for hybrid styles (e.g., character + art style).
• Use the same seed to compare different prompts or LoRA combinations.
• If your LoRA uses trigger words, include them in your prompt.

Notes

• LoRAs can be stacked for combined effects.
• For best results, use LoRAs trained on FLUX-compatible models.
• 9B model offers better detail than 4B at slightly higher cost.

Related Models

• FLUX.2 Klein 9B Text-to-Image — Standard version without LoRA support.
• FLUX.2 Klein 4B Text-to-Image LoRA — Lighter 4B version at lower cost.
• FLUX.2 Pro — Flagship-quality text-to-image generation.

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",
  "size": "1024*1024"
}
JSON
)

# 1. Submit the prediction once. Never retry this request automatically:
# a retry after an unclear failure can create a second, billed task.
CHECK='The task may already exist. Check History or the predictions list before submitting again.'
if ! SUBMIT_RESPONSE=$(curl --silent --show-error --fail-with-body \
  --connect-timeout 10 --max-time 60 \
  -X POST "https://api.wavespeed.ai/api/v3/wavespeed-ai/flux-2-klein-9b/text-to-image-lora" \
  -H "Authorization: Bearer ${WAVESPEED_API_KEY}" \
  -H "Content-Type: application/json" \
  -d "${REQUEST_BODY}"); then
  printf 'Submit failed. %s\n' "${CHECK}" >&2
  exit 1
fi

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 "${PREDICTION_ID}" ]; then
  printf 'Submission response did not contain a prediction id. %s\n' "${CHECK}" >&2
  exit 1
fi
RESULT_URL="https://api.wavespeed.ai/api/v3/predictions/${PREDICTION_ID}/result"

# 2. Poll until the prediction reaches a final status.
BODY_FILE=$(mktemp)
trap 'rm -f "${BODY_FILE}"' EXIT
while true; do
  # Transient errors (429, 5xx) are retried, honoring Retry-After.
  HTTP_CODE=$(curl --silent --show-error --fail-with-body \
    --connect-timeout 10 --max-time 60 --retry 5 --retry-delay 2 \
    --output "${BODY_FILE}" --write-out '%{http_code}' \
    "${RESULT_URL}" \
    -H "Authorization: Bearer ${WAVESPEED_API_KEY}" || true)
  case "${HTTP_CODE}" in
    2??) ;;
    000|429|5??)  # dropped connection or temporary error: keep polling
      printf 'Task %s: HTTP %s, polling again in 10s\n' "${PREDICTION_ID}" "${HTTP_CODE}" >&2
      sleep 10; continue ;;
    *) printf 'Task %s: HTTP %s\n' "${PREDICTION_ID}" "${HTTP_CODE}" >&2
      cat "${BODY_FILE}" >&2; exit 1 ;;
  esac
  RESULT=$(jq 'if type == "object" and has("data") then .data else . end' "${BODY_FILE}")
  STATUS=$(printf '%s' "${RESULT}" | jq -r '.status // empty')

  case "${STATUS}" in
    completed) printf '%s\n' "${RESULT}" | jq '.outputs'; break ;;
    failed|cancelled|timeout|deleted)
      printf 'Task %s ended with status %s\n' "${PREDICTION_ID}" "${STATUS}" >&2
      printf '%s\n' "${RESULT}" | jq . >&2; exit 1 ;;
    *) sleep 2 ;;
  esac
done

Parameters

Task Submission Parameters

Request Parameters

Response Parameters

Result Request Parameters

Result Response Parameters