From 42a7e469c45ae023ad54ebdd90665c37fc66afb1 Mon Sep 17 00:00:00 2001
From: Tony Lin <tony.lin@intel.com>
Date: Thu, 22 Oct 2026 05:13:48 +0000
Subject: [PATCH] docs: correct NIXL hybrid SSM heterogeneous TP support

Signed-off-by: Tony Lin <tony.lin@intel.com>
---
 docs/features/nixl_connector_compatibility.md | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/docs/features/nixl_connector_compatibility.md b/docs/features/nixl_connector_compatibility.md
index 46cb0bc482c4..05084e58966b 100644
--- a/docs/features/nixl_connector_compatibility.md
+++ b/docs/features/nixl_connector_compatibility.md
@@ -52,7 +52,7 @@ th:not(:first-child) {
 | Dense Transformers | ✅ | ✅<sup>1</sup> | ✅ | ✅<sup>2</sup> | ✅ | ✅ | 🟠<sup>3</sup> |
 | MLA (e.g. DeepSeek-V2/V3) | ✅ | ✅<sup>1</sup> | 🟠<sup>4</sup> | ✅<sup>2</sup> | ✅ | ✅ | 🟠<sup>3</sup> |
 | Sparse MLA (e.g. DeepSeek-V3.2) | ✅ | ✅<sup>1</sup> | 🟠<sup>4</sup> | ✅<sup>2</sup> | ✅ | ✅ | 🟠<sup>3</sup> |
-| Hybrid SSM / Mamba | ✅ | ❔ | 🚧<sup>5</sup> | ❌ | ✅ | ✅ | ❌<sup>6</sup> |
+| Hybrid SSM / Mamba | ✅ | ❔ | 🟠<sup>5</sup> | ❌ | ✅ | ✅ | ❌<sup>6</sup> |
 | MoE | ✅ | ✅<sup>1</sup> | ✅ | ✅<sup>2</sup> | ✅ | ✅ | 🟠<sup>3</sup> |
 | Multimodal | ❔ | ❔ | ❔ | ❔ | ❔ | ❔ | ❔ |
 | Encoder-Decoder | ❌ | ❌ | ❌ | ❌ | ❌ | ❌ | ❌ |
@@ -67,7 +67,7 @@ what may differ.
 
 <sup>4</sup> MLA KV cache is replicated across TP workers, so heterogeneous TP works but there is no head-splitting. When P TP > D TP, only a single read is executed (redundant ranks are skipped). D TP > P TP also works.
 
-<sup>5</sup> Hybrid SSM (Mamba) models require **homogeneous TP** (`P TP == D TP`). Heterogeneous TP is not yet supported for Mamba layers.
+<sup>5</sup> Hybrid SSM/Mamba models (including GDN) support heterogeneous TP in both directions, provided one TP size is an integer multiple of the other.
 
 <sup>6</sup> HMA (required by hybrid models) does not support different remote block sizes.