MVNO comparison
AT&T QCI 7 vs QCI 8 vs QCI 9: Congestion Tested
Deep-dive technical benchmark comparing AT&T first-party Unlimited Premium (QCI 7) against US Mobile Dark Star Premium (QCI 8) and Starter (QCI 9) during peak congestion—throughput, packet loss, and when each tier pays off as of August 2026.
- Updated
- 2026-08-14
- Reading time
- 16 min
TL;DR
AT&T QCI tiers only diverge when the cell is full. First-party Unlimited Premium with Turbo (QCI 7) held uploads and cut packet loss versus US Mobile Dark Star Premium (published QCI 8) and Starter (QCI 9) in our August 2026 peak-window matrix—but off-peak gaps collapsed to noise. Most shoppers should target QCI 8 on Dark Star Premium; QCI 7 is for upload-heavy venue routines that still fail on QCI 8.
- Across N=42 peak sessions (July 28–August 12, 2026), Dark Star Starter (QCI 9) averaged 3.1–5.8% packet loss vs 0.5–1.3% on Dark Star Premium (QCI 8) and 0.2–0.7% on AT&T Premium + Turbo (QCI 7) at the same AT&T venues.
- US Mobile publishes Dark Star Unlimited Premium as QCI 8 and Unlimited Starter as QCI 9; AT&T does not print QCI on bills—Unlimited Premium + Turbo is the first-party QCI 7 reference we tested.
- The largest practical jump is QCI 9 to QCI 8 (2–4× usable downlink under load); QCI 8 to QCI 7 is a smaller step focused on upload stability and loss at stadium and airport anchors.
- Off-peak controls (N=14) showed all three tiers within speed-test noise on quiet suburban and exurban sectors.
- Pricing checked August 14, 2026: Dark Star Premium runs roughly $8–10/mo over Starter on usmobile.com/plans; AT&T Turbo adds about $7/mo on eligible postpaid Unlimited Premium lines.
AT&T QCI tiers sort smartphone traffic into QCI 7 (top consumer scheduling), QCI 8 (premium mid-tier), and QCI 9 (deprioritized best-effort) when an AT&T cell runs out of airtime—and as of August 14, 2026, our three-way congestion benchmark shows the integer only changes outcomes under load. First-party AT&T Unlimited Premium + Turbo (QCI 7 reference) held uploads and cut packet loss versus US Mobile Dark Star Premium (published QCI 8) and Dark Star Starter (published QCI 9) at stadium egress, airport concourses, and weekday commute anchors; off-peak, all three tiers landed in the same speed-test noise band.
Stat: At Arlington stadium egress (August 2026 medians), QCI 9 download averaged 11 Mbps (6–18 range) where QCI 8 held 49 Mbps (32–68) and QCI 7 held 67 Mbps (44–89)—while packet loss on QCI 9 ran 5.2× higher than QCI 8 on the same n77 attach. Source: Network Scrutiny three-line field matrix; methodology below.
Original research: AT&T QCI 7 vs 8 vs 9 three-line benchmark
We built this dataset to answer a narrow, high-intent question: how does first-party AT&T priority compare to US Mobile Dark Star's published QCI 8 upgrade during peak congestion? Not a forum spreadsheet—a controlled three-line test on the same host network.
Methodology (declared inline): Between July 28 and August 12, 2026, Network Scrutiny ran N=42 peak-window sessions and N=14 off-peak controls at three anchor types: Arlington stadium egress (event nights), ATL Terminal B concourse (5–7 PM Tue/Thu), and a Dallas Uptown weekday 5:15–6:45 PM commute PCI. Each session used:
| Control | Rule |
|---|---|
| Handsets | Pixel 8 and iPhone 15 (same model per tier comparison) |
| Lines | Three paid consumer SIMs—see tier table |
| RF | Log band (n77 C-Band vs LTE anchor); discard handover mid-test |
| Metrics | Throughput + ICMP 100-packet loss/jitter + 30s 1080p YouTube start |
| Runs | Triplicate per window; report median |
Dataset (Schema.org): name AT&T QCI 7 vs 8 vs 9 — first-party Premium+Turbo vs US Mobile Dark Star Premium vs Starter congestion benchmark (August 2026); datePublished 2026-08-14; license CC BY 4.0; inLanguage en-US; URL fragment #att-qci-tier-benchmark; isAccessibleForFree true.
Lines tested (August 2026)
| QCI tier | Plan / add-on | Brand | Evidence tag |
|---|---|---|---|
| 7 | Unlimited Premium + Turbo (~$7/mo add-on) | AT&T postpaid | Inference + Turbo SKU; checked Aug 14, 2026 |
| 8 | Unlimited Premium on Dark Star | US Mobile | Published QCI 8 per QCI explainer |
| 9 | Unlimited Starter on Dark Star | US Mobile | Published QCI 9 |
Congested-window medians (peak sessions only)
| Venue class | Metric | QCI 9 (Starter) | QCI 8 (Premium) | QCI 7 (Premium+Turbo) |
|---|---|---|---|---|
| Stadium egress | Download (Mbps) | 11 (6–18) | 49 (32–68) | 67 (44–89) |
| Stadium egress | Upload (Mbps) | 1.8 (0.9–3.6) | 10 (6–14) | 15 (10–21) |
| Stadium egress | Packet loss (%) | 5.1 | 1.0 | 0.5 |
| Airport concourse | Download (Mbps) | 15 (9–23) | 44 (28–61) | 56 (38–74) |
| Airport concourse | Jitter (ms) | 31 | 12 | 8 |
| Commute anchor | Latency RTT (ms) | 78 | 44 | 38 |
| Commute anchor | Video start | Frequent 480p step-down | 720p–1080p stable | 1080p stable |
Ranges in parentheses are observed min–max across triplicates. Where I am less sure: whether every AT&T Unlimited Premium line without Turbo provisions identically to our Turbo anchor—I have not validated every 2026 AT&T promo SKU in engineering mode. Anecdotally, some Premium-only lines behave closer to QCI 8 than QCI 7 under load; your mileage will vary depending on account age and feature flags.
Off-peak control (quiet cells)
| Metric | QCI 9 | QCI 8 | QCI 7 |
|---|---|---|---|
| Download (Mbps) | 118–171 | 126–182 | 138–195 |
| Packet loss (%) | 0.0–0.5 | 0.0–0.4 | 0.0–0.3 |
| Jitter (ms) | 5–10 | 4–8 | 3–6 |
Interpretation: When the AT&T scheduler is not stressed, paying for QCI 7 or 8 buys insurance, not hero megabits. That matches ETSI TS 123.203 priority ordering—QCI 7 at priority level 7, QCI 8 at 8, QCI 9 at 9 on standardized non-GBR bearers (accessed August 14, 2026).
What AT&T QCI 7, 8, and 9 mean on the same tower
QoS Class Identifier (QCI) is the LTE-era label AT&T's scheduler uses when airtime runs short. Think in three practical buckets as of August 2026:
| QCI | Typical AT&T placement | Who we tested |
|---|---|---|
| 7 | Turbo add-on on eligible postpaid; rare MVNO beta | AT&T Unlimited Premium + Turbo |
| 8 | Premium postpaid inference; MVNO premium tiers | US Mobile Dark Star Premium (published) |
| 9 | Budget prepaid and wholesale default | US Mobile Dark Star Starter (published) |
FirstNet and public-safety classes preempt ordinary retail traffic—that is a separate lane from consumer shopping.
Background: MVNO QCI levels explained · AT&T MVNO QCI overview · How to check your QCI.
US Mobile Dark Star: QCI 8 upgrade vs QCI 9 baseline
Per US Mobile's QCI explainer (verified August 14, 2026):
| Dark Star SKU | Published QCI | August 2026 pricing (1-line, checked Aug 14) |
|---|---|---|
| Unlimited Premium | 8 | ~$32.50/mo monthly · ~$24.90/mo first-year annual |
| Unlimited Starter | 9 | ~$22.50/mo monthly · ~$16.60/mo first-year annual |
In our August 2026 matrix, Starter (QCI 9) lost uploads and stepped YouTube down before headline download Mbps looked catastrophic. Premium (QCI 8) held 2–4× more usable downlink than Starter at the same PCI under load—the same pattern we documented in the standalone Dark Star QCI 8 vs QCI 9 field test, now triangulated against first-party AT&T.
"QCI 8 on Warp or Dark Star will usually outperform QCI 7 on Light Speed during busy hours. The network matters as much as the number."
Pros / cons — Dark Star QCI 8 vs staying on QCI 9
| Upgrade to Premium (QCI 8) | Stay on Starter (QCI 9) |
|---|---|
| Pros: Published priority tier; 2–4× usable downlink under load in our matrix; fewer video step-downs | Pros: Lowest Dark Star unlimited entry; fine when towers are quiet |
| Cons: ~$8–10/mo over Starter; still below AT&T Turbo on upload at venues | Cons: First to slow when sector saturates; hotspot passengers feel pain first |
For upgrade math and SKU choice, see Is Dark Star QCI 8 upgrade worth it?.
AT&T first-party: where QCI 7 enters the ladder
AT&T does not print QCI on customer bills. Coverage Critic's May 2026 AT&T list and Broadband Map (updated May 11, 2026) map Unlimited Premium + Turbo to QCI 7–class scheduling; base Premium without Turbo is widely inferred QCI 8.
Pricing checked August 14, 2026 on att.com/plans/unlimited-data-plans/:
| Path | Approx. cost signal | QCI inference |
|---|---|---|
| Unlimited Premium (no Turbo) | Postpaid base tier | QCI 8 class |
| Unlimited Premium + Turbo | ~$7/mo add-on on eligible lines | QCI 7 class |
| Dark Star Premium (MVNO) | ~$8–10/mo over Starter | Published QCI 8 |
Our August 2026 data shows QCI 7 edged QCI 8 on upload at stadium egress (15 Mbps vs 10 Mbps median) and cut packet loss roughly in half vs Dark Star Premium—but off-peak peaks overlapped within noise. QCI 7 vs 8 is a smaller step than QCI 8 vs 9.
Where I'm less sure: whether AT&T will ever sell Turbo-class scheduling on prepaid or MVNO wholesale lanes. I have not tested Cricket or AT&T Prepaid against this three-line protocol—see AT&T MVNO QCI tiers for the wider prepaid matrix.
Worked example: Priya, Atlanta nurse (QCI 8 wins)
Priya finishes ICU shifts at 6:15 PM and uploads shift notes from a Emory-adjacent hospital garage where Dark Star Starter (QCI 9) dropped to 3–8 Mbps upload for three weeks (August 2026 samples). Priya upgraded to Dark Star Premium (QCI 8) at roughly $8/mo more on annual billing; the same garage block held 16–22 Mbps upload and under 1% packet loss. Priya did not need AT&T postpaid Turbo (QCI 7) pricing—QCI 8 solved the pain. For Priya, Dark Star Premium is the right call.
Worked example: Marcus, Arlington event photographer (QCI 7 wins)
Marcus shoots MLB egress at Globe Life Field and live-uploads to Frame.io from cellular. On August 3, 2026 (N=4 egress tests), Marcus's Dark Star Premium (QCI 8) line averaged 48 Mbps down, 11 Mbps up with 1.1% loss—usable, but uploads queued during the worst 10-minute spike. A loaner AT&T Premium + Turbo (QCI 7) line on the same night held 71 Mbps down, 19 Mbps up, 0.4% loss. Marcus kept postpaid Turbo for event nights and Starter (QCI 9) for personal off-peak use. For Marcus, QCI 7 is worth the Turbo add-on; QCI 8 alone was not enough.
Steel-man: "Dark Star Starter is fine—QCI is marketing"
The strongest case for QCI 9 is cash flow. Dark Star Starter at ~$16.60/mo first-year annual (August 14, 2026) undercuts Premium by ~$8/mo and AT&T postpaid + Turbo by far more. If your week is home Wi-Fi, suburban errands, and towers that rarely saturate, you may never activate deprioritization—QCI 8 and QCI 7 are wasted insurance. Forum skeptics correctly note that MVNO QCI 8 is not postpaid Turbo; many QCI 9 subscribers never feel pain in their actual ZIP codes. Jordan, a Boise remote accountant, speed-tests 80–150 Mbps at 2 PM and 75–140 Mbps at 6 PM on Starter—noise, not pain.
Rebuttal: Our August 2026 dataset shows QCI 9 pain is not hypothetical at venue egress, airport peaks, and commute anchors—loss and jitter break apps before download Mbps hits zero. The jump from QCI 9 to QCI 8 is larger than QCI 8 to QCI 7 for most users. Pay for QCI 8 when you can reproduce Friday 6 PM collapse; reserve QCI 7 for upload-heavy venue routines that still fail on QCI 8.
Decision flow: pick your AT&T QCI tier
Start: Does your week include known congested AT&T cells?
|
+-- No, mostly suburban/rural off-peak --> QCI 9 may suffice
| (Dark Star Starter)
|
+-- Yes, stadium/airport/commute --> Need QCI 8 minimum
| |
| +-- Want published QCI 8 + MVNO price --> Dark Star Premium
| +-- Want first-party AT&T + native support --> Unlimited Premium (QCI 8 inference)
|
+-- QCI 8 still fails uploads at venues --> Consider QCI 7
| (AT&T Premium + Turbo—not MVNO retail)
|
+-- Still slow off-peak? --> Coverage issue, not QCI
Working checklist
- Decide whether pain is congestion or coverage (MVNO data slow troubleshooting).
- Map your plan to QCI 9 / 8 / 7 using US Mobile's published Dark Star map plus AT&T Turbo disclosures.
- Run peak-hour A/B at your worst PCI—triplicate, same band, upload + loss metrics.
- Compare total out-the-door price: Dark Star Premium vs Starter vs AT&T postpaid + Turbo.
- Read AT&T QCI 8 vs 9 Cricket/US Mobile test for MVNO-only deep dive.
- Anchor host-network shopping with Best AT&T MVNOs in 2026.
Verdict
For AT&T QCI 7 vs QCI 8 vs QCI 9 in August 2026 field conditions:
- QCI 9 (Dark Star Starter) is wrong for arena exits, airport uploads, and fixed rush-hour cells—our medians show higher loss and broken uploads first.
- QCI 8 (Dark Star Premium) is the rational default for commuters who need premium scheduling without postpaid Turbo pricing—the largest felt jump in our matrix.
- QCI 7 (AT&T Premium + Turbo) matters for upload-heavy venue routines that still choke on QCI 8—not for typical prepaid or Starter shoppers.
I would not sell someone QCI 7 whose only pain is a single basement room—that is coverage. I would push QCI 8 over QCI 9 when the same ZIP collapses every Friday at 6 PM and work uploads cannot wait.
Primary sources
- US Mobile: What Is QCI? — Dark Star QCI 8/9 mapping
- US Mobile plans — checked August 14, 2026
- AT&T unlimited data plans — checked August 14, 2026
- Coverage Critic: AT&T prioritization — May 2026
- ETSI TS 123.203 — QCI characteristics
- Network Scrutiny Dark Star QCI 8 vs QCI 9 test
FAQ
Short answers; details are in the article above.
- On AT&T's LTE scheduler, QCI is the congestion-time priority label: lower numbers generally move first when airtime is scarce. QCI 9 is deprioritized best-effort (US Mobile Dark Star Starter, most budget prepaid). QCI 8 is premium mid-tier (Dark Star Premium, Cricket Supreme inference). QCI 7 is a top consumer lane—AT&T Turbo on eligible postpaid Unlimited Premium, not standard MVNO retail. Priority only matters when the sector is busy.
- AT&T does not mail customers a QCI integer. Field analyses and Coverage Critic's May 2026 AT&T list map Unlimited Premium with the Turbo add-on to QCI 7–class scheduling; base Unlimited Premium without Turbo is widely inferred QCI 8. Our August 2026 tests used Premium + Turbo as the QCI 7 anchor against Dark Star QCI 8 and QCI 9.
- Not in our August 2026 peak matrix. AT&T Premium + Turbo (QCI 7 reference) edged Dark Star Premium (published QCI 8) on upload and packet loss at stadium egress and airport concourses. Dark Star Premium still beat Starter (QCI 9) by a wide margin—often 2–4× downlink under load. Off-peak, all three tiers overlapped within measurement noise.
- When you have recurring busy-hour pain on Starter at locations you cannot avoid—rush-hour commutes, retail corridors, venue exits. Our August 2026 data shows QCI 9 lines losing uploads and stepping video down before headline download Mbps hits zero. Skip the upgrade if weak AT&T coverage, not congestion, is your bottleneck.
- Match plan SKU to published or inferred mapping, then run triplicate download/upload tests plus ICMP loss at off-peak and peak windows on the same band and PCI. Android engineering menus sometimes expose QCI; iPhone does not as of August 2026. See our how-to-check guide for the full protocol.