Summary

  • Arcep published more than 465,000 mobile-service measurements from Guadeloupe, Martinique, French Guiana, Saint-Barthélemy and Saint-Martin, collected between May and July 2026.
  • Results from Réunion and Mayotte are still due; the 2026 campaign also tested 5G in Guadeloupe and Martinique for the first time.

The map has five territories on it and two still in the waiting room. On 22 October, France’s telecom regulator published the Antilles–Guyane results of its 2026 mobile-service campaign: more than 465,000 measurements taken from May to July. Réunion and Mayotte will get a later release. Arcep supplied no publication date for that second batch.

That is a useful public record with a visible edge. It tells residents, local authorities and network planners what the regulator measured, where it measured it and which everyday jobs a mobile connection could handle. It does not yet give the seven overseas territories one common, current snapshot.

The tests were built around tasks rather than a single speed trophy. Arcep checked whether calls lasted two minutes without audible disruption, scored voice quality, timed SMS delivery against a ten-second mark, measured upload and download speeds, timed web pages and assessed video quality. Its download thresholds—3, 8 and 30 Mbit/s—are tied to different uses, from ordinary browsing to video and demanding workplace tools. A network is more than the largest number it can flash on a chart.

The 2026 campaign also changed what its instruments could see. Arcep used 5G-compatible test chains for the whole campaign and for the first time in Guadeloupe and Martinique. Terminals could move freely across 2G, 3G, 4G and 5G. The regulator also measured calls made through internet messaging apps in 10% of the tests, alongside ordinary voice calls.

That wider toolkit makes the new survey more useful, and it complicates year-to-year comparisons. A new radio technology entering the test path changes the experience being measured. A 2025 result and a 2026 result can sit side by side only when the geography, task and connection mode are clear. Otherwise, the chart may look like a trend while the measuring tape has changed.

There is a second distinction in the release. A field test records what a connection did at a place and time. A coverage map starts with modeled predictions based on network declarations. Arcep checked the reliability of newly enriched 4G maps and says some local corrections followed. Those are different evidence layers: one is a forecast, the other a measurement. A buyer or public planner needs to know which one is under the cursor.

The geographic gap is plain. Five territories have results now; Réunion and Mayotte do not. The missing fact is the date and measurement window for those two results. Until they arrive, a national or cross-territory procurement comparison built from this release has an obvious hole, even if the Antilles–Guyane sample is large.

Arcep has made the tests public and the underlying campaign data available through its open-data catalogue. The next useful move is simple: publish the remaining territories with their dates, scope and method beside the first five. A benchmark earns trust territory by territory; the map cannot fill in the last two by optimism.

Sources