Home Expressions
Docs
Rust Docs QAIL PG (Rust) Zig Docs QAIL Zig
Drivers Gateway SDKs Benchmarks
Changelog
Rust / Workspace qail.rs and workspace releases Zig qail-zig release track
GitHub
Rust / Workspace github.com/qail-io/qail Zig github.com/qail-io/qail-zig
Blog Status Roadmap
Performance Metrics

Transparent Benchmarks & Methodology

A meticulously tracked index of QAIL runtime benchmarks across PostgreSQL, Qdrant, ORMs, and IPC surfaces. Verify workload shapes, data topologies, and repro commands before you quote any result.

Verified Results qail.rs qail-zig Reproducible Builds

How to read this index

Dated snapshots

Every report is a point-in-time measurement. Quote the report date, workload shape, and runtime assumptions with the number.

Subsystem-specific

PostgreSQL driver, Zig runtime, Qdrant, ORM, and IPC results are tracked separately so one benchmark does not stand in for the whole platform.

Reproducible first

Canonical pages should explain harness, dataset, and reproduction commands before making performance claims.

Current Reports

These are the current benchmark reports to use when evaluating an active subsystem.

March 27, 2026
Current Zig report

qail-zig PostgreSQL driver

Pure Zig driver/runtime vs qail.rs and pgx
Three-way isolated report for prepared single, prepared pipeline, and pool10 modes. This is the canonical benchmark surface for qail-zig after the Rust/Zig split.
zigpostgresqldriver
January 1, 2026
Current Qdrant report

qail-qdrant vector driver

Single-query, pool, and HTTP/2 batch search
Qdrant-specific report focused on transport shape, batching, and client comparison for the vector driver path.
qdranthttp/2vector
December 31, 2025
Current ORM snapshot

qail-pg vs SQLx vs SeaORM

50K complex-query comparison on the Rust PostgreSQL path
Matched warmup, connection setup, and complex workload for a technical driver-vs-ORM comparison rather than a homepage headline number.
ormpostgresqlrust

Transport Reports

These reports document cross-language and IPC transport work without conflating them with the core driver pages.

December 27, 2025
Transport report

Go IPC vs pgx

50M-query prepared workload over the IPC daemon path
Focused on the Go transport boundary, prepared statements, and the cost of the daemon split compared with direct pgx execution.
goipcpostgresql
December 27, 2025
Transport report

Python IPC vs PyO3

10M-query Python transport comparison
Documents the socket-based Python path separately from Rust-native results so transport conclusions stay grounded in the right workload.
pythonipctransport

Archived Reports

Older reports stay published for context, but they should be read as archive material unless a newer current report does not exist.

December 27, 2025
Historical language snapshot

Cross-language performance

PHP, Go, and Python implementation snapshots
Archive page covering multiple language tracks. Useful for history, but not the canonical benchmark surface for any single runtime today.
phpgopython
December 27, 2025
Historical Zig page

PostgreSQL Zig 50M query run

Earlier Zig benchmark surface vs pg.zig
Retained for search and historical comparison. The current Zig benchmark read lives in /zig/docs/benchmarks.html.
zigarchivepostgresql
December 2025
Historical ORM page

ORM/driver benchmark

Earlier CTE and JOIN comparison page
Predates the 50K ORM snapshot and remains useful as a dated technical note rather than the default benchmark page.
ormarchivecte
December 26, 2025
Historical COPY snapshot

COPY and query throughput

Bulk insert and query throughput snapshot
Production-scale COPY and query throughput results kept online as archive material with their original assumptions.
copypostgresqlarchive
December 23, 2025
Historical pipeline snapshot

Pipelining throughput

Early sequential vs pipelined comparison
Useful for understanding the evolution of the pipeline story, but not the page to use as the current canonical benchmark summary.
pipelinearchivethroughput

Reproduce locally

Local reruns matter more than copied headline numbers. Use the repository and harness that matches the report you are reading, then confirm hardware, data, and network assumptions from that report page.

qail.rs workspace

Use the target report page for dataset shape, connection topology, and any required environment variables.

cargo run --example orm_benchmark --releasecargo run --example fair_benchmark --releasecargo run --example batch_benchmark --release Open repository

qail-zig workspace

The current public Zig benchmark methodology is documented in /zig/docs/benchmarks.html and backed by the qail-zig repository.

zig build ast-benchzig build bench Open repository