FLUX.2-dev

A 32 billion parameter rectified flow transformer designed for image generation, editing, and combination based on text instructions. It supports open-ended tasks such as text-to-image generation, single-reference editing, and multi-reference editing without requiring additional finetuning. Trained using guidance distillation to enhance efficiency, the model is optimized for research and creative applications under a non-commercial license.

Key Features:

• State-of-the-art performance in text-to-image generation, single-reference editing, and multi-reference editing.
• No finetuning needed for character, object, or style references. Read more in the guide.
• Trained with guidance distillation to improve computational efficiency.
• Open weights for scientific research and artistic workflow development.
• Outputs can be used for personal, scientific, or commercial purposes under the FLUX [dev] Non-Commercial License.
• Deployable via the `diffusers` library with quantization support (e.g., 4-bit) for consumer GPUs like RTX 4090/5090.
• Pre-training data filtered for NSFW content and CSAM using partnerships with the Internet Watch Foundation. Post-training safety fine-tuning to suppress unlawful content generation (e.g., CSAM, NCII). Third-party evaluations and iterative fine-tuning ensured resilience against adversarial prompts and reference images. Inference filters for NSFW/IP-infringing content, pixel-layer watermarking, and C2PA metadata integration for provenance.
• Licensing terms prohibit unlawful use, with policies enforced via developer agreements and monitoring. Commercial use requires separate licensing. 

---

The model is a component of the image generation pipeline, consisting of:

• Text encoder: ~24B parameters,
• Transformer: ~32B parameters,
• VAE: ~84M parameters.

Total: ~56B parameters

Announce Date: 18.11.2025
Parameters: 32B
Context: 512
VRAM requirements: 14.4 GB using 4 bits quantization, 28.8 GB using 8 bits quantization, 57.7 GB using 16 bits quantization
Developer: Black Forest Labs
License: FLUX [dev] Non-Commercial License v2.0

Use our pre-built public endpoints for free to test inference and explore FLUX.2-dev capabilities. You can obtain an API access token on the token management page after registration and verification.

Rent your own physically dedicated instance with hourly or long-term monthly billing.

We recommend deploying private instances in the following scenarios:

• maximize endpoint performance,
• enable full context for long sequences,
• ensure top-tier security for data processing in an isolated, dedicated environment,
• use custom weights, such as fine-tuned models or LoRA adapters.

Contact our dedicated neural networks support team at nn@immers.cloud or send your request to the sales department at sale@immers.cloud.