Quote the paper title now that it is not a link

index.html

@@ -528,7 +528,7 @@ html{scroll-behavior:smooth}
     </article>
     <article class="key-win">
       <h3>latency reaches the first token</h3>
-      <p>The teams at Crusoe and NVIDIA make this case especially well in <a href="https://www.crusoe.ai/resources/blog/reducing-ttft-by-cpumaxxing-tokenization">Reducing TTFT by CPUMaxxing Tokenization</a>. Every prompt must be tokenized before inference can return its first token. That cost becomes visible in time to first token for long agent contexts and requests that reuse a cached model prefix.</p>
+      <p>The teams at Crusoe and NVIDIA make this case especially well in "<a href="https://www.crusoe.ai/resources/blog/reducing-ttft-by-cpumaxxing-tokenization">Reducing TTFT by CPUMaxxing Tokenization</a>." Every prompt must be tokenized before inference can return its first token. That cost becomes visible in time to first token for long agent contexts and requests that reuse a cached model prefix.</p>
       <div class="mini-chart" id="keyLatencyChart"></div>
       <p class="note">Median p99 across eight model families, measured call by call on an Apple M4 Max. Lower is better.</p>
     </article>