diff --git a/README.md b/README.md
index 79e923616..4c148133a 100644
--- a/README.md
+++ b/README.md
@@ -34,9 +34,12 @@ Aider is unique in that it [works well with pre-existing, larger codebases](http
 ## New GPT-4 model with 128k context window
 
 Aider supports OpenAI's new GPT-4 model that has the massive 128k context window.
-[Early benchmark results](https://aider.chat/docs/benchmarks-1106.html)
-indicate that it is very fast and
-a bit better at coding than previous GPT-4 models.
+[Early benchmark results]
+indicate that it is
+[very fast](https://aider.chat/docs/benchmarks-speed-1106.html).
+and a bit
+[better at coding](https://aider.chat/docs/benchmarks-1106.html).
+than previous GPT-4 models.
 
 To use it, run aider like this:
 
diff --git a/docs/benchmarks-speed-1106.md b/docs/benchmarks-speed-1106.md
index cc974f086..f14b5353d 100644
--- a/docs/benchmarks-speed-1106.md
+++ b/docs/benchmarks-speed-1106.md
@@ -40,8 +40,6 @@ Some observations:
 - **GPT-4 Turbo is 4-5x faster.** The new `gpt-4-1106-preview` model is 4-5x faster than the June (0613) version which has been the default `gpt-4` model.
 - The old March (0301) version of GPT-3.5 is actually faster than the June (0613) version. This was a surprising discovery.
 
-### Preliminary results
-
 **These are preliminary results.**
 OpenAI is enforcing very low
 rate limits on the new GPT-4 model.