STAC Insights

Independent Research. Expert Driven. Always Relevant.

STAC Insights does what it says, provides insights to technical topics and data.  If Benchmarks answer the 'what' then Insights provides the 'how' and 'why'.

STAC Insights

Setting the standard for thought leadership in financial technology. Our studies go surface trends, delivering deep, data-driven insights that shape strategic decisions. By engaging with a global network of innovators and practitioners, we provide clarity on the technologies that will define the future of finance.

Contact us to discuss how STAC Insights might benefit your firm.

What we research

STAC Research examines the full spectrum of emerging technologies and strategies in capital markets, delivering actionable insights into how firms innovate and adapt. Our research spans critical domains such as advanced analytics, low-latency systems, HPC, AI and machine learning, cloud and edge computing, and next-generation infrastructure. Each study addresses real-world challenges, helping firms align technology choices with business goals.

We take a question-led approach, exploring the how, why, and what if behind key issues, and occasionally testing hypotheses to validate ideas.

Restricted Reports are exclusive content, often as a means of providing early access to new products that may not be considered General Available.  These reports are produced by the technology vendors utilising STAC Benchmarks tools but without the independent validation our audited STAC Reports provide. Keep that in mind when interpreting results but these often present very interesting early results.

Insights also offers deeper visibility into performance data and system configurations featured in STAC Reports. This level of detail helps you understand performance characteristics and the exact environment setup—so you can focus on the metrics that matter most to you.

Latest INSIGHTS

STAC Report
Posted: October 28, 2025
STAC-AI™ LANG6 on HPE ProLiant DL380a and NVIDIA H200 GPUs
STAC Report
Posted: October 28, 2025
STAC-AI™ LANG6 on HPE ProLiant DL384 and NVIDIA GH200 Superchips
STAC Report
Posted: October 27, 2025
STAC-M3 Benchmark Results: KX kdb+ 4.1 on Intel, Supermicro, Micron,
Unaudited Report
Posted: October 24, 2025
STAC-A3 Unaudited report by Curated Time Technology
STAC Report
Posted: October 13, 2025
STAC-ML™ Markets (Inference) new world record set by NVIDIA and Supermicro
STAC-ML™ Markets (Inference) on a NVIDIA GH200 Grace Hopper Superchip in a Supermicro server
Research Note
Posted: August 6, 2025
LLM-Based RAG Evaluation Metrics
LLM-Based RAG Evaluation Metrics: Model Relatedness and Consistency
Unaudited Report
Posted: July 29, 2025
STAC-AI™ LANG6 on NVIDIA GB200 Grace Hopper Superchip
NVIDIA publish unaudited STAC-AI inferencing benchmark results for GB200
Unaudited Report
Posted: July 2, 2025
STAC-AI™ LANG6 on NVIDIA GH200 Grace Hopper Superchip
NVIDIA publish unaudited STAC-AI inferencing benchmark results for GH200
Unaudited Report
Posted: May 19, 2025
STAC-A2 Risk Computation on 2x Intel 6980P Processors with RDIMMs
Intel recently performed the STAC-A2 Benchmark tests on a stack consisting of 2 x Intel Xeon 6980P P
STAC Report
Posted: May 13, 2025
STAC-A2 Pack for oneAPI (Rev R) with 2 x Intel Xeon 6980P Processors, Micron MRDIMMs and Red Hat Enterprise Linux 9.5
STAC recently performed STAC-A2 Benchmark tests on a stack consisting of 2 x Intel Xeon 6980P Proces
Research Note
Posted: May 8, 2025
Comparing LLM Benchmarking Frameworks
We recently conducted a study comparing multiple LLM benchmarking frameworks, including the STAC-AI™
Research Note
Posted: April 20, 2025
Performance And Efficiency Comparison Between Self-Hosted LLMs And API Services
We recently conducted a study comparing self-hosted LLMs and equivalent API models using the STAC-AI
STAC Report
Posted: April 6, 2025
Extending STAC-ML with Gradient Boosted Tree Models
STAC has completed a Proof of Concept (POC) benchmark, extending the STAC-ML™ Markets (Inference) to
Research Note
Posted: February 13, 2025
LLM Model Serving Platform Comparison
We recently conducted a study comparing two model-serving platforms - vLLM and Hugging Face’s TGI
Unaudited Report
Posted: December 19, 2024
STAC-AI™ LANG6 on NVIDIA GB200 Grace Hopper Superchip
NVIDIA publish unaudited STAC-AI inferencing benchmark results for GB200

Why it matters

Community-led

STAC research reflects the collective knowledge and real-world priorities of 600+ member firms.

Unbiased & Independent

All benchmarks & research are developed & conducted and tested free from vendor bias.

decision-ready

Our research & reports give firms the clarity they need to compare technologies, justify investments, and move faster.

built for practitioners

From infrastructure engineers to AI leads to CTO's, STAC empowers technologists and decision makers with evidence they can trust.

Sign up to
our newsletter