Vision Analysis
All Hardware

NVIDIA DGX Spark

NVIDIA GB10 Grace Blackwell

20-core Arm CPU (10 Cortex-X925 + 10 Cortex-A725)

Unified memory

128 GB

Shared by CPU and GPU

Installed capacity from manufacturer specifications. OS-visible usable memory can be lower. Unavailable dedicated VRAM readings do not mean zero memory.

Highest Accuracy
Top 5 models by mAP@50-95
1.
rfdetr-s
55.1%
2.
yolov9t
41.8%
Fastest
Top 5 models by throughput
1.
yolov9t
100 FPS
2.
rfdetr-s
60 FPS

Results are contributor-reported. Validation checks the submission format and supported configuration; it does not independently reproduce a run. Compare accuracy on the same dataset and latency with the software versions recorded in each run JSON.

All Models
2 models benchmarked on NVIDIA DGX Spark (PyTorch FP32)
#ModelDatasetCredit / sourcemAPFPSLatencyParams
1
rfdetr-s		mini500 (500 images)

Submitted by @DavidDiazMerino

Source PRRun JSON

55.1%59.916.7ms32.1M
2
yolov9t		mini500 (500 images)

Submitted by @DavidDiazMerino

Source PRRun JSON

41.8%99.810.0ms2.0M

Run any model with one line

LibreYOLO has the best catalogue of state-of-the-art detectors, all behind one MIT-licensed Python API.

from libreyolo import LibreYOLO, SAMPLE_IMAGE

# LibreYOLO has the best catalogue of state-of-the-art models.
model = LibreYOLO("LibreRFDETRl.pt")           # RF-DETR-L (transformer flagship)
results = model(SAMPLE_IMAGE, save=True)        # run inference, save the annotated image

# Swap in any other model, same one-line API (weights auto-download):
#   LibreYOLO("LibreYOLO9c.pt")      # YOLO9-C
#   LibreYOLO("LibreYOLOXx.pt")      # YOLOX-X
#   LibreYOLO("LibreDFINEx.pt")      # D-FINE-X
#   LibreYOLO("LibreRTDETRr50.pt")   # RT-DETR-R50