diff --git a/docs/plans/2026-03-11-cuda-backtest-gpu-residency-plan.md b/docs/plans/2026-03-11-cuda-backtest-gpu-residency-plan.md
index 27b089e99..f5a165154 100644
--- a/docs/plans/2026-03-11-cuda-backtest-gpu-residency-plan.md
+++ b/docs/plans/2026-03-11-cuda-backtest-gpu-residency-plan.md
@@ -721,9 +721,12 @@ fn compute_q_diagnostics_gpu(
 
     // Batch all gap stats into a single tensor to minimize readbacks:
     // [mean_gap, min_gap, max_gap] — one to_vec1 instead of three to_scalar
-    let mean_gap = gaps.mean_all()?;           // [1]
-    let min_gap = gaps.min(0)?.0.min(0)?.0;    // scalar tensor
-    let max_gap = gaps.max(0)?.0.max(0)?.0;    // scalar tensor
+    // Note: In candle-core (git 671de1d), min(D)/max(D) return Result<Tensor>,
+    // NOT Result<(Tensor, Tensor)>. Flatten first for scalar reduction.
+    let gaps_flat = gaps.flatten_all()?;
+    let mean_gap = gaps_flat.mean_all()?;      // scalar tensor
+    let min_gap = gaps_flat.min(0)?;           // scalar tensor
+    let max_gap = gaps_flat.max(0)?;           // scalar tensor
     let gap_stats = Tensor::cat(
         &[&mean_gap.unsqueeze(0)?, &min_gap.unsqueeze(0)?, &max_gap.unsqueeze(0)?], 0
     )?;