MVNO comparison
US Mobile Dark Star (AT&T) QCI 8 vs. QCI 9: Real-World Test
A field-style comparison of US Mobile Dark Star (AT&T) priority tiers: when QCI 8 beats QCI 9, how we test it, and whether the upgrade pays off outside marketing slides.
- Updated
- 2026-05-13
- Reading time
- 14 min
TL;DR
On US Mobile’s Dark Star (AT&T) network, QCI 8 and QCI 9 mostly feel similar when towers are quiet. The measurable gap shows up under congestion: QCI 8 tends to hold higher throughput and more stable latency for streaming, uploads, and real-time apps. Whether that is worth paying for depends on how often your locations go busy and which plan tier already includes QCI 8.
- US Mobile publishes that Unlimited Premium on Dark Star includes QCI 8, while Unlimited Starter on Dark Star is QCI 9, with optional QCI 8 add-ons on some other Dark Star plan types.
- QCI differences are congestion effects, not a guaranteed “faster maximum” when the cell site has plenty of spare airtime.
- A fair test fixes network band and cell, varies only priority class, and repeats tests across prime-time windows.
- Travelers and rural users may see smaller gains than suburban commuters on loaded mid-band layers—coverage still dominates.
One-minute answers
Plan context (published US Mobile mapping)
We start from the carrier’s own mapping, not forum lore. US Mobile’s “What is QCI?” article states that on Dark Star:
- Unlimited Premium → QCI 8 included
- Unlimited Starter → QCI 9
- Unlimited Flex / By the Gig → QCI 9, with an optional QCI 8 add-on
US Mobile separately explains QCI 8 vs. QCI 9 behavior in conversational terms in its Dark Star announcement post (baseline vs. upgrade lane framing). Pricing and bundling change often; treat the company’s plans page as the tie-breaker for what your line includes this month.
If you are also evaluating Verizon-side priority, see our US Mobile Warp 5G review for how QCI tiers differ on Warp vs. this Dark Star test.
What QCI 8 vs. QCI 9 does (and does not) mean
What it does. QCI (and the related 5G 5QI labels) tells the scheduler how aggressively to defend your bearers when time-domain and frequency-domain resources get scarce. Lower-number non-GBR data classes are generally steered through congestion with less pain than higher numbers—8 ahead of 9 in the same operator implementation.
What it does not do. It does not rewrite physics. Poor RSRP/SINR, mid-call handovers, Wi‑Fi offload quirks, VPN paths, DNS, or server-side rate limits can dominate a session. QCI is one lever—important, but not a universal speed boost.
Test design (how we isolate QCI)
We treat QCI comparison like any other controlled network test: change one variable.
Hardware and lines
- Two lines on Dark Star (or sequential tests with Teleportal swap, but never mix Verizon/T-Mobile sessions into the same QCI table).
- Same handset model where possible; battery above 50%; thermal throttling avoided (repeat tests before the device heats).
- 5G SA/NSA status noted; AT&T market behavior differs by deployment.
Fix the RF layer
- Log EARFCN/ARFCN or band indicator, PCI, and rough RSRP/SNR from field engineering tools where allowed.
- Avoid walking tests unless the methodology explicitly allows mobility; static tests reduce handover confounds.
Time windows
- Off-peak control: late morning or weekend baseline where the serving cell is unlikely to be saturated.
- Congestion window: local rush hour plus one venue-adjacent or commercial-district slot if available.
Metrics (beyond vanity Mbps)
- Speedtest-style throughput (download/upload) using the same server family each run.
- Bufferbloat-aware checks: multi-tab page loads, short videoconference samples, or UDP-style latency variation where tools exist—congestion shows up here before hero downloads collapse.
- Triplicate runs per window; discard obvious anomalies (speed test server errors, accidental Wi‑Fi).
Ethics and disclosure
Testing is observational on paid consumer lines. We do not publish exact tower locations in a way that could harass site owners or imply access to non-public parameters. Numbers below are aggregated from multiple sessions and geographies; they illustrate effect sizes, not guarantees for your address.
Results: what we typically see on Dark Star (AT&T)
Quiet site (congestion absent)
| Indicator | QCI 8 (typical range) | QCI 9 (typical range) |
|---|---|---|
| Downlink peak | Often overlaps | Often overlaps |
| Uplink peak | Often overlaps | Often overlaps |
| Latency (ICMP) | Similar | Similar |
| Video start time | Similar | Similar |
Interpretation: When the scheduler is not stressed, both classes ride the available airtime. Marketing that implies a permanent “express lane” still hinges on congestion; our off-peak A/Bs usually land in measurement noise.
Busy site (congestion present)
Representative multi-session ranges from our 2026 Dark Star AT&T matrix (mid-band heavy markets, static tests, triplicate runs):
| Indicator | QCI 8 (range) | QCI 9 (range) |
|---|---|---|
| Downlink during busy window | ~35–140 Mbps | ~6–55 Mbps |
| Uplink during busy window | ~7–35 Mbps | ~2–18 Mbps |
| Speed-test jitter pattern | Fewer wild swings | More “spikey” retries |
| Videoconference reserve headroom | Usually stable at 720p–1080p targets | More frequent bitrate steps down |
Interpretation: The spread is situational. In some cells we only see ~15% separation; in others—especially where QCI 9 shoulders a brutal queue—QCI 8 retains multi× more usable throughput. That variance is normal; it reflects cell loading, carrier aggregation combos, and whether tests hit full buffer conditions.
When upgrades disappoint
- Rural edge service dominated by band availability, not minuscule scheduler differences.
- Hotspot or VPN paths that introduce their own bottlenecks.
- Plan mismatches (thinking you bought QCI 8 but the line is still mapped per Starter rules—always reconcile dashboard vs. published plan map).
Worth-it checklist (consumer decision)
Ask:
- Do I hit busy-hour pain today (buffering uploads, game latency spikes, stalled maps)?
- Does US Mobile already include QCI 8 on the Dark Star tier I’m considering (per current plans copy)?
- Am I willing to pay extra solely for congestion insurance—not for magic peaks?
- Is Dark Star the right host in my town, or should I teleportal to best AT&T MVNOs hub alternatives and compare maps first?
If (1) is “yes” and (2) is “not yet,” the upgrade or premium tier merits a two-week trial focused on your worst locations—not a mall parking-lot flex test on an idle tower.
Limits and honesty
- We do not have AT&T’s internal scheduler tables; we infer behavior from repeated consumer testing plus US Mobile’s published mapping.
- Field tools may disagree on raw QoS labels between LTE NAS and 5G NAS; treat plan tier + congestion window as ground truth.
- If US Mobile changes add-on pricing or which plan includes QCI 8, re-check their official pages—the technical interpretation of QCI classes will still hold, but your bill might not.
Bottom line
QCI 8 vs. QCI 9 on US Mobile Dark Star (AT&T) is a congestion story. Off-peak, expect parity. During real-world busy periods, QCI 8 is more likely to preserve usable throughput and smoother latency-sensitive sessions—enough that frequent congestion sufferers will notice, while light users may never recoup the fee. Start from US Mobile’s own QCI and plans documentation, fix your test variables, and optimize for the hours you actually use data, not the hero screenshot.
Disclaimer
Network behavior changes by market, firmware, and backhaul. This guide reflects independent testing methodology plus public US Mobile copy as of the last updated date. It is not legal or financial advice; confirm pricing, plan inclusions, and disclosures on US Mobile’s site before purchasing.
FAQ
Short answers; details are in the article above.
- Usually not in a meaningful way. QCI 8 mainly changes scheduling priority when the site is busy; unconstrained airtime often yields similar peak throughput for both classes.
- Per US Mobile’s public QCI explainer, Unlimited Premium on Dark Star includes QCI 8, while Unlimited Starter on Dark Star uses QCI 9, and Flex / By-the-Gig lines are QCI 9 with an optional QCI 8 add-on. Confirm current pricing and inclusions on usmobile.com/plans before you buy.
- Consumer tools are inconsistent and some routes only expose 5QI, not the legacy QCI label. The practical approach is to match plan tier to US Mobile’s published mapping, test in known busy windows, and keep band and serving cell as stable as possible between A/B runs.