diff --git a/docs/benchmarks.md b/docs/benchmarks.md
index 3e733a9..6542b3c 100644
--- a/docs/benchmarks.md
+++ b/docs/benchmarks.md
@@ -113,15 +113,20 @@ Same machine, same afternoon: `zig build bench -Doptimize=ReleaseSafe` here,
 
 Context that keeps these honest:
 
-- Upstream jetstream V2 vendors its own decode — no indigo dependency — so
-  the atproto-bench indigo numbers say nothing about it. The zat-vs-their-Go
-  comparison on *format* work is the block-decode row (2.4x) via the same
-  sealed-segment shape (`atproto-bench/go-jss` uses their `segment` package).
+- Upstream jetstream V2 ingests through atmos (`jcalabro/atmos v0.2.14`),
+  which atproto-bench measures directly against zat. The SDK gap is
+  path-dependent: CID-verified decode is zat 249K vs atmos 91K frames/s
+  (~2.7x), but the combined relay hot path — decode + CID verify + commit
+  signature verify — is zat 19.7K vs atmos 18.1K (~1.1x), because ECDSA is
+  ~95% of the frame budget and every implementation pays it. On the
+  signature-verifying ingest path the language advantage mostly washes out;
+  it survives on the sig-free paths (format work, serving, scans), which is
+  what the table above measures — the zat-vs-their-Go format comparison is
+  the block-decode row (2.4x), via `atproto-bench/go-jss` driving their
+  `segment` package on the same sealed-segment shape.
 - The live network runs ~200-450 events/s. Every row above is in the
   millions/s: CPU is not the live tail's bottleneck for either
-  implementation. Where the margin actually pays: bootstrap decode+verify
-  (SDK-level: zat 249K frames/s CID-verified vs 91K for the fastest Go SDK
-  measured in atproto-bench), replay serving, compaction/rebuild scans, and
-  box size for the same workload.
+  implementation. Where the margin actually pays: replay serving, cold
+  fan-out, compaction/rebuild scans, and box size for the same workload.
 - Nothing here is a promotion gate; parity rows marked NOT clean need a
   matched-shape bench before being quoted as a ratio.