diff --git a/memory/.dream_cursor b/memory/.dream_cursor
index fb32aea..8c91667 100644
--- a/memory/.dream_cursor
+++ b/memory/.dream_cursor
@@ -1 +1 @@
-296
\ No newline at end of file
+298
\ No newline at end of file
diff --git a/memory/MEMORY.md b/memory/MEMORY.md
index b02158f..daa7bdc 100644
--- a/memory/MEMORY.md
+++ b/memory/MEMORY.md
@@ -44,6 +44,11 @@ This file stores important information that should persist across sessions.
 - Primary nanobot agent model: GLM-5.1:cloud (best sustained productivity over long sessions/hundreds of rounds, strong real-world agent benchmarks)
 - Alternative nanobot agent model for tool-heavy tasks: Qwen 3.5:cloud (397B variant)
 - Conservative fallback nanobot agent model: DeepSeek V3.2:cloud
+- OpenRouter is pay-per-token alternative to Ollama subscription for model access
+- Gemini Flash via Google AI Studio Free Tier: 15 RPM limit — unusable for agent work; only viable for simple prompts without tool calls
+- Gemini Flash via Google AI Studio Tier 1 (paid): 360+ RPM
+- Haiku (Claude) via OpenRouter: high rate limits, viable agent model alternative
+- Gemini Flash Lite via OpenRouter: high rate limits, viable agent model alternative
 
 ## Reminder System