diff --git a/backends/v3/src/lib.rs b/backends/v3/src/lib.rs
index 77a9a11a..af66b21e 100644
--- a/backends/v3/src/lib.rs
+++ b/backends/v3/src/lib.rs
@@ -100,6 +100,7 @@ pub async fn connect_backend(
             .map_err(V3Error::Warmup)?,
     )?;
     tracing::info!("Setting max batch total tokens to {max_batch_total_tokens}");
+    metrics::gauge!("tgi_batch_max_total_tokens").set(max_batch_total_tokens);
 
     let backend_info = BackendInfo {
         waiting_served_ratio,
diff --git a/router/src/server.rs b/router/src/server.rs
index a6b69321..cc896f99 100644
--- a/router/src/server.rs
+++ b/router/src/server.rs
@@ -1993,8 +1993,6 @@ async fn start(
         "Batch size of the next batch"
     );
 
-    metrics::counter!("tgi_max_token_capacity").increment(u64::from(max_batch_total_tokens));
-
     // CORS layer
     let allow_origin = allow_origin.unwrap_or(AllowOrigin::any());
     let cors_layer = CorsLayer::new()