PC Guide is reader-supported. When you buy through links on our site, we may earn an affiliate commission. Read More
Nvidia has announced its new image-editing model for RTX AI PCs. FLUX.1 Kontext is now available as an Nvidia NIM microservice. The model is powered by Nvidia RTX and TensorRT. Specifically, it’s the [dev] version of the model that is available to RTX AI PC users. Nvidia talks about the new version in its latest blog post, boasting improved performance and much lower VRAM usage.
This news comes less than a month after AMD released its new Amuse 3.1 AI image generation in partnership with Tensorstack. Amuse 3.1 also enjoyed reduced memory usage, dropping the requirement to just 9GB on a 24GB laptop. Nvidia does you one better by dropping it as low as 7GB for FP4 (Blackwell architecture).
FLUX.1 Kontext [dev] performance is 2x faster and uses less VRAM
When compared to the older BF16 PyTorch model, FP8 (Ada Lovelace generation GPUs) and FP4 (Blackwell generation GPUs) offer twice the performance, with the newer Blackwell GPUs pulling ahead. Nvidia also demonstrates that VRAM usage has been cut in half for the Ada Lovelace GPUs when compared to BF16, down from 24GB to 12GB, or as low as 7GB on Blackwell.
Source: Nvidia
The FP4 checkpoint is optimized for the RTX 50 series GPUs on Blackwell architecture, while Ada Lovelace refers to the architecture used on RTX 40 series cards, which are optimized for FP8. As you’d expect, the newer Blackwell cards benefit the most from the new AI model. The flagship RTX 5090 boasts a massive 32GB of VRAM, and as we’ve seen in recent tests versus the cut-down RTX 5050D v2 for the Chinese market, the extra VRAM makes a difference for AI workloads.
Prime Day may have closed its doors, but that hasn’t stopped great deals from landing on the web’s biggest online retailer. Here are all the best last chance savings from this year’s Prime event.
*Prices and savings subject to change. Click through to get the current prices.
These dramatic performance gains were previously limited to AI specialists and developers with advanced AI infrastructure knowledge. With the FLUX.1 Kontext [dev] NIM microservice, even enthusiasts can achieve these time savings with greater performance.
Nvidia Blog
