Bad data doesn't announce itself.
It shows up weeks later, in a survey you already delivered. Drop an observation file β six seconds, no install, no account β and know whether it's usable before you build anything on it.
See how the verdict is calculatedQuick check, no account
One observation file: is this data worth processing, and what would stop it? You get a verdict, the score and up to three findings. The full report, the download and the next step need a free account.
One file up to 100 MB (20 MB gzipped), ten checks per day. The file is deleted as soon as the result is ready.
A result you can inspect, not a promise.
The bundled BJFS sample scored 69.6 in Fast Mode: all epochs were present, but no multipath measurements were recorded, earning 0/30 for that component. Its sampling interval is 30 seconds. Open the bundled example in the free quality check for the latest computed result.
How this verdict is calculatedThree ways to finish the job.
Format conversion
Turn receiver logs into clean RINEX observations without maintaining vendor utilities.
Convert a file
Precise positioning
Run PPP, PPK, or a network solution with inputs, reference frame, and deliverables kept together.
Choose a solution
Coordinate transformation
Move a point or a batch into the datum and projection your deliverable actually requires.
Transform coordinatesConfidence should have a second opinion.
Precise positioning can be cross-checked by two independent scientific-grade geodetic engines. Agreement is measured point by point instead of inferred from one engine's confidence.
Here's how often we fail.
These published 30-day completion rates include files the engines could not solve. They are a dated operating snapshot, not a guarantee.
Live reliability readings load when you open this page.
Success rate = completed Γ· (completed + failed). Only retained tasks submitted in the last 30 days are counted; queued and running tasks are excluded. Restarts replace earlier attempts, and deleted tasks are absent. Median time covers completed tasks with valid start/end times, excluding queue time. Readings are cached for five minutes; reload for a newer snapshot. Conversion and coordinate jobs are not included.
Infrastructure failure is refunded. Credit-hold positioning failures release their reserved credits. Legacy runs refund recognized infrastructure failures and pre-compute input rejections, but keep other charges. See the full boundaries and timing below.
Result files are deleted. Free results are kept 30 days, Pro results 90 days, and Business results 365 days. Cleanup runs daily at 04:45 UTC. The stored outputs are removed on that schedule. The task itself stays in your history as an expired entry, so you can see what ran and when.
Receiver inputs
Data standards
Reference systems
Frequently Asked Questions
What is PosFlow?
PosFlow is a cloud-native GNSS data processing platform. It integrates RINEX format conversion (Flux), quality control (Crux), high-precision positioning (Apex), and coordinate transformation (Axis) into a single web workspace.
Which GNSS constellations and formats are supported?
PosFlow supports all 5 global and regional constellations: GPS, GLONASS, Galileo, BeiDou, and QZSS. Supported formats include RINEX 2.x, 3.x, 4.x, Hatanaka compressed (.crx/.##d), gzip, and raw receiver logs from u-blox (UBX), Septentrio (SBF), Trimble, Leica, Topcon, NovAtel, and Javad.
How does PosFlow compare to traditional desktop GNSS tools?
Unlike legacy desktop software, PosFlow runs cloud-native multi-GNSS precision analysis and geodetic positioning engines in parallel with zero local installation, automated ephemeris pairing, and unified interactive reports.
Is there a free tier available?
Yes. PosFlow provides a permanent Free plan with 25 credits per month β five single-file Fast Mode analyses, or one Full Mode analysis β a 100 MB file limit, all four QC modes, and full downloadable reports. No credit card required.