Most developers building latency-sensitive applications have had to accept a frustrating constraint: the network is not theirs to control. They can optimize every layer of their application stack, but the moment data hits a congested mobile cell, a surgeon loses haptic precision, a maintenance technician sees their AR overlay freeze, or a remote driver gets half a second of lag. The CAMARA project – an open-source initiative under the Linux Foundation and backed by the GSMA Open Gateway program – is trying to change that. The Quality on Demand (QoD) API is its clearest practical expression.
What CAMARA and QoD Actually Are
CAMARA began in 2021, co-founded by Deutsche Telekom, Ericsson, and others. Its goal is to define open, standardized APIs that expose mobile network capabilities to developers – abstracting away the complexity of individual operator implementations so the same code works across networks. The commercial momentum behind it is significant: 86 operator groups representing over 300 networks worldwide, covering around 80% of global mobile connections, have committed to the GSMA Open Gateway program as of early 2026.
The QoD API reached its r3.2 release in Fall 2025 and now covers three interrelated specifications: Quality-on-Demand, QoS Profiles, and QoS Provisioning. In plain terms: a developer makes an API call, specifies a QoS profile – stable low latency, higher throughput, or both – for a defined device and data flow, and the operator’s network handles fulfillment. The developer gets a session ID and a webhook callback if conditions change. No telecom engineering required. No operator-specific integration code.
The Latency Problem in AR/VR Is a Physiology Problem
Human perception of visual lag has a hard limit. Research consistently puts the motion-to-photon threshold at around 20 milliseconds for VR and mixed reality. Above that, the disconnect between physical movement and visual response produces the sensory mismatch that causes cybersickness – nausea, disorientation, headaches. For AR specifically, high latency causes digital overlays to drift from their physical anchor points, breaking the spatial relationship the application depends on.
Industrial AR is where this becomes a commercial problem. Siemens Energy has been one of the most prominent early adopters of QoD. Using Deutsche Telekom and T-Mobile’s QoD API via Microsoft Azure Programmable Connectivity, Siemens Energy’s field technicians use Microsoft HoloLens 2 headsets running Remote Assist, with remote engineers guiding them through maintenance tasks in real time. Without QoD, video quality degrades under network load and the AR overlay becomes unreliable. With it, the network prioritizes that specific data flow on demand – triggered by the application itself.
Remote Surgery: Where Latency Has a Different Kind of Consequence
Telesurgery pushes latency requirements into a territory where the stakes are not commercial but clinical. A 2025 systematic review in the Journal of Robotic Surgery puts the safe operating threshold at under 200 milliseconds, with real-world Chinese 5G-enabled human procedures reporting round-trip latency of 52-159 ms at distances between 1,000 and 5,000 km. The first Telesurgery Consensus Conference in Orlando in February 2024 – attended by over 200 surgeons, the FDA, and every major robotic surgery company – is the source cited by the clinical literature for establishing this sub-200 ms consensus.
The 2024 demonstration connected a surgeon in Orlando to Dubai over 10,000 km – a live telesurgery procedure enabled by 5G, though at around 500 ms latency due to distance. Subsequent Orlando-to-Shanghai procedures at the same event used animal models as a feasibility proof. Both were enabled by Ericsson’s 5G network capabilities – demonstrating that the network architecture for intercontinental surgical connectivity exists. QoD APIs built on 5G Standalone networks are programmable interfaces that make this accessible to healthcare developers without the need for bespoke operator agreements. The architecture is there; the standardized API layer is what CAMARA is building out.
| PRO TIP: Build and Test the Fallback Before the Happy Path The QoD API sends a webhook callback when the network can no longer fulfill the requested QoS profile – status: NETWORK_TERMINATED. Most developers test under good network conditions and assume it works. The critical question is what your application does when that callback fires mid-session: does it degrade gracefully, alert the user, or fail silently? For safety-critical applications like surgical robotics or remote driving, the fallback path is as important as the integration itself. Build and test it before reaching any operator sandbox. |
Remote Driving: A Use Case Already in Production
The furthest-along real-world QoD implementation right now is not in a hospital – it is in a car park. Elmo, an Estonian company operating road-legal remotely driven vehicles, deployed QoD via Nokia’s Network as Code platform with Finnish operator Elisa in spring 2024, remotely driving a car through public streets in Espoo. The QoD API, triggered by a button on the remote driver’s console, immediately prioritized bandwidth and latency for the vehicle’s multi-camera video stream – the same stream the driver depends on to brake safely. Nokia and Hrvatski Telekom launched a similar pilot in Croatia in October 2024.
What makes Elmo significant is that it demonstrates QoD’s practical value in a production-grade safety context – and that the integration was achieved through Nokia’s developer-friendly SDK without custom operator negotiations. The Matsuko holographic meeting platform, cited by T-Mobile DevEdge, used QoD to improve latency, reduce jitter, and increase bandwidth for real-time hologram quality. These are not lab demonstrations; they are functioning integrations across commercial 5G networks.
QoD Adoption: Where It’s Being Used Today
QoD deployments are no longer confined to lab pilots. Commercial and production implementations are live across several countries, with a growing roster of enterprises and operators actively using the API:
- China – China Telecom’s QoD API has exceeded 5 billion cumulative calls as of 2024, with Tencent Games, Bosch, Honor, and Huawei Cloud among its enterprise users. Use cases span online gaming, online education, connected vehicles, and autonomous driving. China Telecom published the world’s first GSMA Open Gateway QoD case study in June 2024.
- Spain – Orange, Telefónica, and Vodafone ran a live QoD showcase on their production networks at MWC 2023, with Blacknut (cloud gaming), Zoom, and Vonage as application partners, focused on high-definition interactive video.
- Germany – Deutsche Telekom, O2 Telefónica, and Vodafone Germany have all commercially deployed QoD, with Deutsche Telekom’s developer portal targeting media companies, automated vehicles, and retail use cases.
- United States – T-Mobile offers QoD directly to developers via its DevEdge portal, with documented integrations including drone operations and race-car telemetry applications. Verizon has been developing a QoS API in partnership with AWS for XR, robotics, and IoT use cases. AT&T, T-Mobile, and Verizon have also joined the Aduna venture. This Ericsson-backed platform combines CAMARA APIs from all three carriers under a unified developer interface, with QoD on the roadmap alongside the initially launched Number Verification and SIM Swap APIs.
The Market Shift Behind All of This
In 2024, security and anti-fraud APIs accounted for over 80% of GSMA Open Gateway commercial launches. By mid-2025, QoD and edge compute services together reached 25% – more than doubling from under 10% the year before. The Fall 2025 CAMARA meta-release delivered 60 standardized network APIs, with QoD designated as a stable, mature API alongside Simple Edge Discovery – a meaningful step toward the “build once, deploy everywhere” goal.
If You’re Building Latency-Sensitive Applications
If you are a developer or system integrator working in healthcare tech, industrial AR, autonomous vehicles, or enterprise XR, QoD is worth evaluating today. Access is available through several routes. Nokia’s Network as Code platform exposes QoD to developers but requires an operator to have deployed it, as Elisa has in Finland and Hrvatski Telekom in Croatia; developers outside those networks need to check whether their target operator has integrated with the platform. For direct developer access, T-Mobile’s DevEdge portal offers CAMARA-compliant QoD APIs on its 5G Standalone network today, and Verizon is building out QoS API capabilities in partnership with AWS. AT&T, T-Mobile, and Verizon are also collectively available through Aduna, the Ericsson-backed aggregation venture. Telefónica Open Gateway and Deutsche Telekom’s developer portal each provide access for European networks, and Microsoft Azure Programmable Connectivity aggregates multiple operators behind a single SDK.
The integration itself is straightforward: a few REST calls against the standardized CAMARA spec, OAuth 2.0 authentication, and a webhook endpoint for status notifications. The harder work is deciding which flows need prioritization, when to trigger a session, and what happens when the network cannot deliver – because for surgical robotics and remote driving, that last question is the one that matters most.
