promrs

Fast Prometheus client for Python, backed by Rust.

promrs is an independent, MIT-licensed reimplementation of the prometheus_client Python API. The core metric logic (counters, gauges, histograms, text encoding) runs in Rust via PyO3, while the Python layer provides the familiar prometheus_client-style interface: Counter, Gauge, Histogram, labels, context managers, and generate_latest.

Why?

	promrs	prometheus_client
License	Apache 2.0	Apache 2.0
Core language	Rust	Python
Output format	OpenMetrics text	Prometheus / OpenMetrics text

• Performance — metric recording and scrape encoding happen in Rust with no GIL overhead. The GIL is released during generate_latest.

Installation

pip install promrs

Requires Python 3.12+. Pre-built wheels are available for Linux (x86_64, aarch64) and macOS (x86_64, Apple Silicon).

Quick start

from promrs import Counter, Histogram, generate_latest, REGISTRY

# Counter
requests = Counter("http_requests_total", "Total HTTP requests", ["method", "path"])
requests.labels(method="GET", path="/api").inc()

# Histogram
latency = Histogram(
    "http_request_duration_seconds",
    "Request latency",
    ["method"],
    buckets=[0.01, 0.05, 0.1, 0.5, 1.0, 5.0],
)
with latency.labels(method="GET").time():
    pass  # timed block

# Scrape
output: bytes = generate_latest()  # uses the default REGISTRY

API

Metric types

Class	Methods	Context managers
Counter	inc(amount=1.0), labels(*args, **kwargs)	count_exceptions(exception=BaseException)
Gauge	set(value), inc(amount=1.0), dec(amount=1.0), set_to_current_time(), labels(...)	time(), track_inprogress()
Histogram	observe(value), labels(...)	time()

All metric constructors accept: name, documentation, labelnames, namespace, subsystem, unit, registry.

Registry

from promrs import CollectorRegistry, Counter, generate_latest

registry = CollectorRegistry()
c = Counter("my_counter", "help", registry=registry)
c.inc()

output = generate_latest(registry)

A default global REGISTRY is provided and used when no registry argument is given.

Content type

from promrs import CONTENT_TYPE_LATEST

# "application/openmetrics-text; version=1.0.0; charset=utf-8"

Development

Prerequisites: Rust, uv.

make deps   # install Python deps + build the Rust extension
make lint   # ruff + pyright + cargo fmt + clippy
make test   # pytest
make bench  # pyperf benchmark vs prometheus_client

License

Apache 2.0