EOSE V11 · GMAT BENCH · Quantitative · Verbal · Data Insights · Day 83 · 2026-04-27
GMAT BENCH 📊
Quant · Verbal · Data Insights · fleet financial data as source material · RICK + MO + SIGNALS lead · γ₁ = 14.134725141734693
0
Correct
0
Wrong
Score %
Est. Score /805
QUANTITATIVE REASONING · Problem Solving + Data Sufficiency
QUANT-001 · QuantSovereign · Problem Solving
The EOSE fleet runs 7 local silos. Monthly GPU compute cost per silo: msi01 = CA$0 (no dedicated GPU cost), msclo = CA$420, yone = CA$380, forge = CA$510, one-deseof = CA$490, pcdev = CA$195, lounge = CA$440. Cloud silos average CA$1,840/month. There are 3 cloud clusters.
If total fleet costs increase by 15% next month and each component increases proportionally, what is the new total monthly cost?
Local silos: 0+420+380+510+490+195+440 = CA$2,435. Cloud: 3×1,840 = CA$5,520. Total = CA$7,279. ×1.15 = CA$8,370.85 ≈ C. Note: April 2026 actual cloud cost tracked at CA$7,332 by day 19 — this problem uses the fleet's real cost architecture.
QUANT-002 · Quantγ₁ Math · Problem Solving
γ₁ = 14.134725141734693. What is the smallest positive integer n such that γ₁ × n > 100?
γ₁×7 = 98.943... < 100. γ₁×8 = 113.077... > 100. Answer: C (n=8). This is a floor-check problem. In FGATE terms: floor_address = γ₁×8 = 113.08 — the first harmonic above LOCO threshold of 100. This is the L5 sovereign threshold anchor.
QUANT-003 · Quant · Data Sufficiency
Is LOCO score L > 80 (L4 maturity level)?
Statement 1: The average of D1 through D5 (critical domains, weight 2x) is 85.
Statement 2: The sum of D6 through D11 (standard domains) is 480.
C. LOCO formula: (2×ΣD1-5 + ΣD6-11) / (2×5 + 6) = (2×5×85 + 480) / 16 = (850+480)/16 = 1330/16 = 83.125 > 80. ✓ But neither statement alone is enough — Stmt 1 only gives D1-5, Stmt 2 only gives D6-11. You need both to compute the weighted average. Together: sufficient. Classic Data Sufficiency: need both, neither alone works.
VERBAL REASONING · Critical Reasoning + Reading Comprehension
VERBAL-001 · Verbal · Critical Reasoning · Strengthen
A study found that companies using sovereign AI deployments (locally-hosted models) reported 23% fewer data breach incidents than companies using third-party cloud AI services over a 12-month period. The study concluded that sovereign AI deployment causally reduces data breach risk.
Which of the following, if true, most strengthens the causal conclusion?
C is correct. To strengthen a causal claim, you want to show temporal sequence (cause precedes effect) or eliminate confounds. C shows that switching TO sovereign deployment CAUSED a subsequent decline — the strongest possible causal evidence (temporal precedence + controlled before/after). A actually weakens slightly — it suggests a confound (better security teams, not deployment type, caused the difference). B is irrelevant — the comparison was with cloud AI, not zero incidents. EOSE: this is exactly the argument structure for our enterprise sales pitch — MOAT-067/078.
DATA INSIGHTS · Table Analysis · Multi-Source Reasoning · MEFINE Data
DATA-001 · Data InsightsMEFINE
Fleet Cost Table — April 2026 (Day 19 snapshot)
ClusterCloudCost CA$ (day 19)GPU TypeProjected Monthly
AKS aks-eose-aaas-devAzure4,200T4 (scaled 0 nights)6,630
AKS pemos-systemAzure1,800None active2,840
ZERO-DR / KRSRHONEGCP NE1820T4/A1001,294
CATHEDRAL / JAYRHONEAWS512V100/A10G808
Total7,33211,572
DATA-001a: If the evening scaledown script reduces the AKS aks-eose-aaas-dev GPU cost by 35% from the projected monthly figure, what is the new projected total monthly fleet cost?
AKS dev projected: 6,630. 35% reduction = 6,630 × 0.65 = 4,309.50. Other clusters unchanged: 2,840 + 1,294 + 808 = 4,942. New total: 4,309.50 + 4,942 = CA$9,251.50. Answer: B. Note: March 2026 actual was CA$10,552 — the scaledown script is already working, bringing April projection from ~11,572 toward the target.