NVIDIA RTX Supercharges FLUX 1 For Next Gen AI Art

A New Era for AI Image Generation with FLUX.1 Kontext

The world of AI image generation just got a major upgrade. Black Forest Labs, a leading AI research lab, has introduced its groundbreaking FLUX.1 Kontext model. This isn't just another image generator; it's a fundamental shift in how creators guide and refine AI-generated visuals.

Previously, AI artists juggled multiple models and complex tools like ControlNets to achieve their desired results. This often meant complex workflows, like using a 3D scene to guide an image's composition, as seen in the NVIDIA AI Blueprint for 3D-guided generation. The new FLUX.1 Kontext model streamlines this entire process, offering a single, powerful tool for both image creation and editing using simple, natural language.

To make this possible, NVIDIA has collaborated with Black Forest Labs to supercharge FLUX.1 Kontext [dev] for NVIDIA RTX GPUs. By leveraging the NVIDIA TensorRT SDK and quantization, the model achieves faster performance with significantly lower VRAM needs. For creators, this means smoother iterations and more direct control, all from their local RTX-powered computers.

Unpacking the Power of FLUX.1 Kontext

Introduced in May, the FLUX.1 Kontext family of models accepts both text and images as prompts. This allows users to start with a reference image and make edits with simple text commands, eliminating the need for fine-tuning or managing multiple ControlNets.

An example of multi-turn editing possible with the FLUX.1 Kontext [dev] model. The original image (left); the first edit transforms it into a Bauhaus style image (middle) and the second edit changes the color style of the image with a pastel palette (right).

The model's key capabilities are designed for creative flexibility:

• Character Consistency: Maintain a character's unique features across different scenes and angles.
• Localized Editing: Change specific parts of an image without affecting the rest.
• Style Transfer: Apply the aesthetic of a reference image to new creations.
• Real-Time Performance: Enjoy low-latency generation for rapid feedback and iteration.

This approach replaces cumbersome editing techniques that required complex masks or depth maps with an intuitive, flexible interface that blends step-by-step editing with powerful performance optimization.

Optimized for RTX with TensorRT Acceleration

NVIDIA and Black Forest Labs worked together to not only simplify the creative workflow but also to make it more accessible. By quantizing the model, they dramatically reduced its memory footprint. The model size was cut from 24GB to 12GB for FP8 precision (ideal for GeForce RTX 40 Series) and down to 7GB for FP4 precision (optimized for the upcoming GeForce RTX 50 Series using a new method called SVDQuant).

On top of the memory savings, TensorRT delivers a massive speed boost, providing over 2x the performance compared to running the original model in PyTorch.

Speedup compared with BF16 GPU (left, higher is better) and memory usage required to run FLUX.1 Kontext [dev] in different precisions (right, lower is better).

How to Get Started with FLUX.1 Kontext

Ready to try it out? The FLUX.1 Kontext [dev] model is available now on Hugging Face, including the TensorRT-accelerated variants.

AI enthusiasts can use the Torch versions in ComfyUI or test the model directly in the Black Forest Labs Playground. An NVIDIA NIM microservice version is also expected in August.

More from the World of RTX AI

There's more exciting news for AI developers and enthusiasts:

• Gemma 3n Accelerated: Google's new multimodal small language model, Gemma 3n, is now accelerated for NVIDIA GeForce RTX GPUs and the NVIDIA Jetson platform. You can learn more about running Gemma 3n on your hardware here.

• G-Assist Hackathon: NVIDIA's Project G-Assist Plug-In Hackathon is ongoing until July 16. Developers are invited to build custom G-Assist plug-ins for a chance to win prizes. Don't miss the informational webinar to get started.

Join the Community

To connect with fellow developers and AI enthusiasts, join NVIDIA’s Discord server for discussions on the latest in RTX AI. You can also stay informed by following the RTX AI PC channels on Facebook, Instagram, TikTok, and X, and by subscribing to the RTX AI PC newsletter.