diff --git a/website/blog/2025-01-22-pure-rl-with-deepseek/deepseek.png b/website/blog/2025-01-22-pure-rl-with-deepseek/deepseek.png
new file mode 100644
index 0000000..a3dc854
Binary files /dev/null and b/website/blog/2025-01-22-pure-rl-with-deepseek/deepseek.png differ
diff --git a/website/blog/2025-01-22-pure-rl-with-deepseek/index.mdx b/website/blog/2025-01-22-pure-rl-with-deepseek/index.mdx
index 04a8b82..342424a 100644
--- a/website/blog/2025-01-22-pure-rl-with-deepseek/index.mdx
+++ b/website/blog/2025-01-22-pure-rl-with-deepseek/index.mdx
@@ -36,6 +36,8 @@ Man, it was pretty good. I've got a smaller GPU on PWS so I was limited to runni
 the best response times on the 1.5b model, and for some easier tasks the correctness between the two was hard to discern.
 
 ### Search
+(This is an update to the original blog post, after experimenting more with search.)  
+
 I particularly liked the features of OpenWebUI to allow for web search, which from initial testing seemed to find good results to build
 context of the response with.