diff --git a/src/services/llm_local.rs b/src/services/llm_local.rs
index 349b3a4..4fd144f 100644
--- a/src/services/llm_local.rs
+++ b/src/services/llm_local.rs
@@ -185,9 +185,13 @@ async fn start_llm_server(
 ) -> Result<(), Box<dyn std::error::Error + Send + Sync>> {
     let port = url.split(':').last().unwrap_or("8081");
 
+    std::env::set_var("OMP_NUM_THREADS", "20");
+    std::env::set_var("OMP_PLACES", "cores");
+    std::env::set_var("OMP_PROC_BIND", "close");
+
     let mut cmd = tokio::process::Command::new("sh");
     cmd.arg("-c").arg(format!(
-        "cd {} && ./llama-server -m {} --host 0.0.0.0 --port {} --threads 8 --threads-batch 18 --temp 0.7 --parallel 3 --repeat-penalty 1.1 --ctx-size 1024 --batch-size 4096 -n 255 --mlock --no-mmap   &",
+        "cd {} && numactl --interleave=all ./llama-server -m {} --host 0.0.0.0 --port {} --threads 20 --threads-batch 40 --temp 0.7 --parallel 1 --repeat-penalty 1.1 --ctx-size 10000 --batch-size 10000 -n 4096 --mlock --no-mmap --prompt-cache-all --flash-attn  --no-kv-offload --tensor-split 100   --no-mmap &",
         llama_cpp_path, model_path, port
     ));