About LibHFT and Atrekes
LibHFT is a low-latency FIX engine family for C++, Java, .NET and Rust, with dated, reproducible benchmark evidence. This page says who builds it, how its evidence is reviewed before it is published, where its source code stands and how to reach the engineers.
Who builds LibHFT
LibHFT is developed by Atrekes, which builds low-latency FIX infrastructure, electronic brokerage platforms and developer tools for teams building and operating electronic-market systems.
This site, libhft.com, carries LibHFT’s documentation, benchmark evidence and the browser-based FIX message parser. The company and its other products live on atrekes.com.
Visit AtrekesHow evidence is reviewed
The rules below are the ones the documentation applies on every page. They decide what may be called a result, what stays labelled as a diagnostic, and what a figure can and cannot be compared with.
A headline figure states its campaign date, host, workload, rate and transport. Published numbers describe the recorded environment, not a latency guarantee on different hardware or topology. How we benchmark a FIX engine.
An adapter must do the same work to enter a comparison. The qualified HP codec campaign of 18 October 2026 checks converted values and complete business-field sets through each public parser API, with 332 isolated text and numeric probes. Parity rules.
Only an S5 pass on all four canonical flows is eligible for a report-grade full-engine comparison. A missed rate, an incomplete cell or a newer functional smoke test supplies no ranked latency, and failed targets stay visible in each language's engine comparison.
Source-symbol coverage, conformance behavior, measured performance and scoped boundaries are kept apart. A generated codec is not a complete venue connector; a raw NIC measurement is not a FIX round trip; a local active/passive test is not cross-datacentre replication. LibHFT documentation.
Results are published with their source revision and SHA-256 hash, and superseded measurement profiles keep their correction notes instead of disappearing. Result catalogue.
The public documentation summarizes committed, conformance-tracked evidence. It is not venue certification or a deployment-specific guarantee: benchmark the flavor, transport and hardware you will operate. Venue evidence boundary.
Source availability
LibHFT’s source code is not public. Evaluation is arranged by contact with the engineering team, for the flavor, transport and workload you need to measure. Any public release will be announced on this site.
Questions the engineers answer directly: evaluating LibHFT, migrating from QuickFIX, kernel-bypass deployments and reproducing a published benchmark.
Contact the LibHFT engineers: contact@atrekes.com