diff --git a/llama.cpp b/llama.cpp
index 99b5c5b39..ec4a7f94c 100644
--- a/llama.cpp
+++ b/llama.cpp
@@ -7359,6 +7359,7 @@ void llama_sample_grammar(struct llama_context * ctx, llama_token_data_array * c
     std::vector<std::pair<std::vector<uint32_t>, llama_partial_utf8>> candidates_decoded;
     candidates_decoded.reserve(candidates->size);
     std::vector<llama_grammar_candidate>                              candidates_grammar;
+    candidates_grammar.reserve(candidates->size);
 
     for (size_t i = 0; i < candidates->size; ++i) {
         const llama_token id    = candidates->data[i].id;