diff --git a/server/download.go b/server/download.go
index 68836a824..5b02991cf 100644
--- a/server/download.go
+++ b/server/download.go
@@ -132,16 +132,8 @@ func (p *blobDownloadPart) UnmarshalJSON(b []byte) error {
 	return nil
 }
 
-// Download tuning. Override via environment variables:
-//   - Memory constrained: reduce OLLAMA_DOWNLOAD_CONCURRENCY
-//   - Default uses ~64KB memory regardless of concurrency (streams to disk, hashes from page cache)
 var (
-	// downloadPartSize is the size of each download part.
-	// Default: 64MB. Override with OLLAMA_DOWNLOAD_PART_SIZE (in MB).
-	downloadPartSize = int64(getEnvInt("OLLAMA_DOWNLOAD_PART_SIZE", 64)) * format.MegaByte
-
-	// downloadConcurrency limits concurrent part downloads.
-	// Default: 32. Override with OLLAMA_DOWNLOAD_CONCURRENCY.
+	downloadPartSize    = int64(getEnvInt("OLLAMA_DOWNLOAD_PART_SIZE", 64)) * format.MegaByte
 	downloadConcurrency = getEnvInt("OLLAMA_DOWNLOAD_CONCURRENCY", 32)
 )
 
@@ -421,6 +413,34 @@ func (b *blobDownload) run(ctx context.Context, requestURL *url.URL, opts *regis
 		close(hashDone)
 	}()
 
+	// Log progress periodically
+	// Page cache warning: if spread > 4GB, hasher may hit disk instead of RAM
+	const pageCacheWarningBytes = 4 << 30 // 4GB
+	progressDone := make(chan struct{})
+	go func() {
+		ticker := time.NewTicker(time.Second)
+		defer ticker.Stop()
+		for {
+			select {
+			case <-ticker.C:
+				downloaded := b.Completed.Load()
+				hashed := sh.Hashed()
+				dlPct := int(downloaded * 100 / b.Total)
+				hPct := int(hashed * 100 / b.Total)
+				spread := dlPct - hPct
+				spreadBytes := downloaded - hashed
+
+				msg := fmt.Sprintf("progress: downloaded %d%% | hashed %d%% | spread %d%%", dlPct, hPct, spread)
+				if spreadBytes > pageCacheWarningBytes {
+					msg += fmt.Sprintf(" [WARNING: %.1fGB ahead, page cache pressure]", float64(spreadBytes)/(1<<30))
+				}
+				slog.Info(msg)
+			case <-progressDone:
+				return
+			}
+		}
+	}()
+
 	g, inner := errgroup.WithContext(ctx)
 	g.SetLimit(downloadConcurrency)
 	for i := range b.Parts {
@@ -462,34 +482,6 @@ func (b *blobDownload) run(ctx context.Context, requestURL *url.URL, opts *regis
 		})
 	}
 
-	// Log progress periodically
-	// Page cache warning: if spread > 1GB, hasher may hit disk instead of RAM
-	const pageCacheWarningBytes = 1 << 30 // 1GB
-	progressDone := make(chan struct{})
-	go func() {
-		ticker := time.NewTicker(time.Second)
-		defer ticker.Stop()
-		for {
-			select {
-			case <-ticker.C:
-				downloaded := b.Completed.Load()
-				hashed := sh.Hashed()
-				dlPct := int(downloaded * 100 / b.Total)
-				hPct := int(hashed * 100 / b.Total)
-				spread := dlPct - hPct
-				spreadBytes := downloaded - hashed
-
-				msg := fmt.Sprintf("progress: downloaded %d%% | hashed %d%% | spread %d%%", dlPct, hPct, spread)
-				if spreadBytes > pageCacheWarningBytes {
-					msg += fmt.Sprintf(" [WARNING: %.1fGB ahead, page cache pressure]", float64(spreadBytes)/(1<<30))
-				}
-				slog.Info(msg)
-			case <-progressDone:
-				return
-			}
-		}
-	}()
-
 	if err := g.Wait(); err != nil {
 		close(progressDone)
 		sh.Stop()