diff --git a/llama.cpp/llama.cpp b/llama.cpp/llama.cpp
index ec58864795..7e92fa9830 100644
--- a/llama.cpp/llama.cpp
+++ b/llama.cpp/llama.cpp
@@ -4321,6 +4321,9 @@ static void llm_load_vocab(
             } else if (
                 tokenizer_pre == "dbrx") {
                 vocab.type_pre = LLAMA_VOCAB_PRE_TYPE_DBRX;
+            } else if (
+                tokenizer_pre == "smollm") {
+                vocab.type_pre = LLAMA_VOCAB_PRE_TYPE_SMOLLM;
             } else {
                 throw std::runtime_error(format("unknown pre-tokenizer type: '%s'", tokenizer_pre.c_str()));
             }
@@ -12185,6 +12188,7 @@ struct llm_tokenizer_bpe {
                     case LLAMA_VOCAB_PRE_TYPE_STARCODER:
                     case LLAMA_VOCAB_PRE_TYPE_REFACT:
                     case LLAMA_VOCAB_PRE_TYPE_COMMAND_R:
+                    case LLAMA_VOCAB_PRE_TYPE_SMOLLM:
                         word_collection = unicode_regex_split(text, {
                             "\\p{N}",
                             "'s|'t|'re|'ve|'m|'ll|'d| ?\\p{L}+| ?\\p{N}+| ?[^\\s\\p{L}\\p{N}]+|\\s+(?!\\S)",
diff --git a/llama.cpp/llama.h b/llama.cpp/llama.h
index 5927f06cdb..78b4890cd0 100644
--- a/llama.cpp/llama.h
+++ b/llama.cpp/llama.h
@@ -87,6 +87,7 @@ extern "C" {
         LLAMA_VOCAB_PRE_TYPE_QWEN2          = 11,
         LLAMA_VOCAB_PRE_TYPE_OLMO           = 12,
         LLAMA_VOCAB_PRE_TYPE_DBRX           = 13,
+        LLAMA_VOCAB_PRE_TYPE_SMOLLM           = 14,
     };
 
     // note: these values should be synchronized with ggml_rope