diff --git a/bindings/python/CHANGELOG.md b/bindings/python/CHANGELOG.md
index 344e8df87..ea8623928 100644
--- a/bindings/python/CHANGELOG.md
+++ b/bindings/python/CHANGELOG.md
@@ -1,6 +1,9 @@
 # v0.6.0 (not published yet)
 
-Fixes:
+## Changes:
+- Big improvements in speed for BPE (Both training and tokenization) ([#165](https://github.com/huggingface/tokenizers/pull/165))
+
+## Fixes:
 - Some default tokens were missing from `BertWordPieceTokenizer` (cf [#160](https://github.com/huggingface/tokenizers/issues/160))
 - There was a bug in ByteLevel PreTokenizer that caused offsets to be wrong if a char got split up
 in multiple bytes. (cf [#156](https://github.com/huggingface/tokenizers/pull/156))
diff --git a/tokenizers/CHANGELOG.md b/tokenizers/CHANGELOG.md
index b0f29a848..31469f36f 100644
--- a/tokenizers/CHANGELOG.md
+++ b/tokenizers/CHANGELOG.md
@@ -1,5 +1,8 @@
 # v0.8.0 (not released yet)
 
+## Changes:
+- Big improvements in speed for BPE (Both training and tokenization) ([#165](https://github.com/huggingface/tokenizers/pull/165))
+
 ## Fixes:
 - Do not open all files directly while training ([#163](https://github.com/huggingface/tokenizers/issues/163))
 - There was a bug in ByteLevel PreTokenizer that caused offsets to be wrong if a char got split up