diff --git a/README.md b/README.md
index 33a98c5f..36e862f6 100644
--- a/README.md
+++ b/README.md
@@ -28,6 +28,12 @@
   <strong>English</strong> · <a href="README.zh-CN.md">简体中文</a>
 </p>
 
+<p align="center">
+  <strong>#1 on every major AI memory benchmark — <a href="https://github.com/xiaowu0162/LongMemEval">LongMemEval</a>, <a href="https://github.com/snap-research/locomo">LoCoMo</a>, and <a href="https://github.com/Salesforce/ConvoMem">ConvoMem</a>.</strong><br/>
+  <strong>95% Recall@15 with a 99.4% context reduction · ~50ms user profiles.</strong><br/>
+  <a href="https://supermemory.ai/research">Read the research →</a>
+</p>
+
 ---
 
 Supermemory is the memory and context layer for AI. **#1 on [LongMemEval](https://github.com/xiaowu0162/LongMemEval), [LoCoMo](https://github.com/snap-research/locomo), and [ConvoMem](https://github.com/Salesforce/ConvoMem)** — the three major benchmarks for AI memory. 
@@ -356,10 +362,14 @@ Supermemory is state of the art across all major AI memory benchmarks:
 
 | Benchmark | What it measures | Result |
 |---|---|---|
-| **[LongMemEval](https://github.com/xiaowu0162/LongMemEval)** | Long-term memory across sessions with knowledge updates | **81.6% — #1** |
+| **[LongMemEval](https://github.com/xiaowu0162/LongMemEval)** | Long-term memory across sessions with knowledge updates | **#1** |
 | **[LoCoMo](https://github.com/snap-research/locomo)** | Fact recall across extended conversations (single-hop, multi-hop, temporal, adversarial) | **#1** |
 | **[ConvoMem](https://github.com/Salesforce/ConvoMem)** | Personalization and preference learning | **#1** |
 
+On LongMemEval, supermemory reaches **95% Recall@15 while adding only ~720 tokens of context — a 99.4% context reduction** (99.6% at @10, 99.8% at @5). Recall by category: Knowledge Updates 99%, Assistant recall 100%, User recall 97%, Multi-session 93%, Temporal Reasoning 91%, Preference 90%.
+
+We also built the **Supermemory Filesystem (SMFS)**, which uses **3.0× fewer tokens on Claude** (24M vs 72M) and **1.75× fewer on Codex** across the 110-question xAFS benchmark. See the full write-ups on our [research page](https://supermemory.ai/research).
+
 We also built **[MemoryBench](https://supermemory.ai/docs/memorybench/overview)** — an open-source framework for standardized, reproducible benchmarks of memory providers. Compare Supermemory, Mem0, Zep, and others head-to-head:
 
 ```bash