Home/Models/Replicate/Black Forest Labs/FLUX 2 DEV
R

Black Forest Labs/FLUX 2 DEV

요청당:$0.06
FLUX 2 DEV는 연구, 실험 및 비상업적 용도에 최적화된 개발자 친화적 버전입니다. 품질과 연산 효율의 균형을 유지하면서 개발자에게 강력한 이미지 생성 기능을 제공합니다. 프로토타이핑, 학술 연구, 개인 창작 프로젝트에 최적입니다.
새로운
상업적 사용
개요
기능
가격
API

Key features (what Flux.2 Dev does)

  • Text→Image generation with high prompt adherence and improved typography / small-detail rendering.
  • Multi-reference editing — combine multiple reference images into a single output, preserving identity/style consistency
  • Single checkpoint for generation + editing (no separate editing model required).
  • Large open-weight checkpoint (32B) permitting local research, quantization, and community adaptation.)
  • Optimized VAE for an improved learnability–quality–compression tradeoff (enables 4MP editing/outputs).

Technical details (architecture & engineering)

  • Parameter count: 32 billion parameters for the FLUX.2 checkpoint.
  • Core design: latent flow-matching / rectified flow transformer combined with a vision-language model (BFL says they couple a Mistral-3 24B VLM with the transformer backbone for semantic grounding). The VLM contributes world knowledge and textual grounding while the transformer models spatial/compositional structure.
  • VAE: new FLUX.2 VAE (released under Apache-2.0) retrained to improve reconstruction fidelity and latent learnability, enabling high-resolution editing.
  • Sampling & distillation: trained using guidance-distillation techniques to improve inference efficiency and fidelity.

Benchmark performance

Black Forest Labs published comparative evaluations and charts showing FLUX.2’s performance vs. contemporary open-weight and hosted image models. Key published figures (BFL / press summaries):

  • Text-to-image win rate: FLUX.2 ~66.6% (vs. Qwen-Image 51.3%, Hunyuan ~48.1% in BFL’s head-to-head dataset).
  • Single-reference editing win rate: FLUX.2 ~59.8% (vs. Qwen-Image 49.3%, FLUX.1 Kontext ~41.2%).
  • Multi-reference editing win rate: FLUX.2 ~63.6% (vs. Qwen-Image 36.4%). BFL also reports multi-reference capability up to 10 references in their evaluation suite.

Typical / recommended use cases

  • Ad and marketing image variants where the same model/actor/product must remain consistent across many scenes or backgrounds (multi-reference consistency).
  • Product photography & virtual try-on (preserve product details across backgrounds).
  • Editorial/fashion spreads requiring the same identity across many shots.
  • Rapid prototyping and research (dev checkpoint allows experimentation, fine-tuning and LoRA/adapter workflows).

How to access Flux.2 dev API

Step 1: Sign Up for API Key

Log in to cometapi.com. If you are not our user yet, please register first. Sign into your CometAPI console. Get the access credential API key of the interface. Click “Add Token” at the API token in the personal center, get the token key: sk-xxxxx and submit.

Flux.2 Dev API

Step 2: Send Requests to Flux.2 dev API

Select the “black-forest-labs/flux-2-dev ”endpoint to send the API request and set the request body. The request method and request body are obtained from our website API doc. Our website also provides Apifox test for your convenience. Replace <YOUR_API_KEY> with your actual CometAPI key from your account.

Insert your question or request into the content field—this is what the model will respond to . Process the API response to get the generated answer.

Step 3: Retrieve and Verify Results

Process the API response to get the generated answer. After processing, the API responds with the task status and output data.

CometAPI Now Supporting Replicate Format Models: 🔹 black-forest-labs/flux-2-pro 🔹 black-forest-labs/flux-2-dev 🔹 black-forest-labs/flux-2-flex

Limited Time Promotion: Lower than Replicate Official Pricing!

👇 Start Building Now Create Predictions – API Doc

⚡ Flexible Selection:

  • Pro: Designed for high-efficiency production and fast delivery.
  • Flex: Maximizes image quality with adjustable parameters.
  • Dev: Developer-friendly optimization.

자주 묻는 질문

Is FLUX.2 [dev] free to use?

Yes, FLUX.2 [dev] is available for free for non-commercial local development and testing purposes.

Can I use FLUX.2 [dev] for commercial projects?

No, FLUX.2 [dev] is strictly licensed for non-commercial use. For commercial applications, you should use the [pro], [flex], or [max] API models.

Where can I download the FLUX.2 [dev] weights?

The weights for FLUX.2 [dev] are available for download on HuggingFace for local inference deployment.

What are the hardware requirements for running FLUX.2 [dev] locally?

While exact requirements vary, local development typically requires a high-performance GPU with significant VRAM (e.g., 24GB+) to handle the model's architecture efficiently.

Does FLUX.2 [dev] support the same features as the API models?

FLUX.2 [dev] allows for full customization and supports core features like text-to-image and basic editing, but lacks cloud-exclusive features like grounding search found in [max].

What is the recommended multi-reference limit for FLUX.2 [dev]?

For local development with FLUX.2 [dev], the documentation recommends using a maximum of 6 reference images, compared to 8-10 supported via the API.

Can I fine-tune FLUX.2 [dev]?

Yes, as an open-weight model, FLUX.2 [dev] serves as a base for community fine-tuning and developing custom LoRAs.

Black Forest Labs/FLUX 2 DEV의 기능

[모델 이름]의 성능과 사용성을 향상시키도록 설계된 주요 기능을 살펴보세요. 이러한 기능이 프로젝트에 어떻게 도움이 되고 사용자 경험을 개선할 수 있는지 알아보세요.

Black Forest Labs/FLUX 2 DEV 가격

[모델명]의 경쟁력 있는 가격을 살펴보세요. 다양한 예산과 사용 요구에 맞게 설계되었습니다. 유연한 요금제로 사용한 만큼만 지불하므로 요구사항이 증가함에 따라 쉽게 확장할 수 있습니다. [모델명]이 비용을 관리 가능한 수준으로 유지하면서 프로젝트를 어떻게 향상시킬 수 있는지 알아보세요.
코멧 가격 (USD / M Tokens)공식 가격 (USD / M Tokens)할인
요청당:$0.06
요청당:$0.075
-20%

Black Forest Labs/FLUX 2 DEV의 샘플 코드 및 API

[모델 이름]의 포괄적인 샘플 코드와 API 리소스에 액세스하여 통합 프로세스를 간소화하세요. 자세한 문서는 단계별 가이드를 제공하여 프로젝트에서 [모델 이름]의 모든 잠재력을 활용할 수 있도록 돕습니다.
Python
JavaScript
Curl
import os
import requests

COMETAPI_KEY = os.environ.get("COMETAPI_KEY") or "<YOUR_COMETAPI_KEY>"

url = "https://api.cometapi.com/replicate/v1/models/black-forest-labs/flux-2-dev/predictions"

headers = {
    "Authorization": f"Bearer {COMETAPI_KEY}",
    "Content-Type": "application/json"
}

payload = {
    "input": {
        "prompt": "A sleek silver sports car racing along a coastal highway at sunset, hyper-realistic, cinematic lighting, 8k",
        "input_images": [
            "https://replicate.delivery/pbxt/O7kbtH7wgLIItlCyeWjm0fgPpx7OpGsT9VbYyxBXnfieVxQe/woman-by-car.jpg"
        ],
        "go_fast": True,
        "aspect_ratio": "16:9",
        "output_format": "jpg",
        "output_quality": 90,
        "seed": 42
    }
}

response = requests.post(url, headers=headers, json=payload)
result = response.json()

print(f"Status Code: {response.status_code}")
print(f"Task ID: {result.get('id')}")
print(f"Status: {result.get('status')}")
print(f"Model: {result.get('model')}")

더 많은 모델