Sub-2µs p99 latency and up to 50 million inferences per second in audited benchmarks unveiled today at STAC Summit London
CAMBRIDGE, England, Oct. 6, 2026 /PRNewswire/ — myrtle.ai today announced that its VOLLO® inference accelerator has set new records on the STAC-ML Markets (Inference) benchmarks for gradient-boosted trees, cutting 99th-percentile latency by more than 30% and raising throughput by at least 5x over the previous best results. The STAC®-audited results were unveiled at the STAC Summit in London today.
Running on an AMD Alveo™ V80LL Compute Accelerator in a Blackcore ICON 3132-SM+ server, VOLLO achieved p99 latencies below 2 microseconds for all three models. For the smallest model, it sustained 50 million inferences per second at a p99 latency of just 1.77µs.
In electronic trading, the time between market data arriving and a decision being made directly affects returns. Low, deterministic latency lets firms run larger, more accurate models without missing the market, so model quality no longer has to be traded off against speed.
“Trading firms want to run ever more powerful models without giving up speed, and these results show they can. Developers can now test their own models on VOLLO without any FPGA expertise and see the difference for themselves,” said Peter Baldwin, CEO, myrtle.ai.
Following the STAC Tacana results announced in April, VOLLO now holds the records for deterministic latency for both decision trees and neural networks. It is already proven in production, with hundreds of thousands of hours of live trading generating alpha for many of the world’s leading trading firms. Its model flexibility has also made it a platform of choice in telecoms, network security and defence.
STAC-ML Markets (Inference) is the technology benchmark standard for running inference on real-time market data. Designed by quants and technologists from leading financial firms, it reports the performance, resource efficiency and quality of any technology stack capable of running the provided models. Full results are in the STAC Report (SUT ID MRTL2026905) at www.STACresearch.com/MRTL2026905.
ML developers can evaluate how their own models would perform on VOLLO today, with no FPGA tools or expertise required. Visit myrtle.ai/vollo-trees or contact [email protected].
About myrtle.ai
Myrtle.ai is an AI/ML software company delivering ultra-low-latency inference accelerators on FPGA-based platforms from all the leading FPGA suppliers. Its accelerators serve applications including financial trading, wireless telecoms, LLMs, speech processing and recommendation.
VOLLO, VOLLO Accelerator and the VOLLO logo are registered trademarks of myrtle.ai. “STAC” and all STAC names are trademarks or registered trademarks of the Strategic Technology Analysis Center, LLC. AMD, the AMD Arrow logo, Alveo, and combinations thereof are trademarks of Advanced Micro Devices, Inc.
View original content to download multimedia:https://www.prnewswire.com/apac/news-releases/myrtleais-vollo-sets-new-stac-ml-records-for-gradient-boosted-tree-inference-302898730.html
SOURCE Myrtle.ai
