diff --git a/backend/cpp/llama-cpp/Makefile b/backend/cpp/llama-cpp/Makefile
index bcd388182..b8b1b08be 100644
--- a/backend/cpp/llama-cpp/Makefile
+++ b/backend/cpp/llama-cpp/Makefile
@@ -1,5 +1,5 @@
 
-LLAMA_VERSION?=a37cc46104eb50900d5ab4a9ba02a0abdac039db
+LLAMA_VERSION?=78a8ae14be559b45f5f13ddee82c188b683d5d8d
 LLAMA_REPO?=https://github.com/ggerganov/llama.cpp
 
 CMAKE_ARGS?=
diff --git a/backend/cpp/llama-cpp/grpc-server.cpp b/backend/cpp/llama-cpp/grpc-server.cpp
index 21e556bef..55f777edb 100644
--- a/backend/cpp/llama-cpp/grpc-server.cpp
+++ b/backend/cpp/llama-cpp/grpc-server.cpp
@@ -107,6 +107,14 @@ using grpc::Status;
 #define LOCALAI_MTMD_INIT_OPT_ARG(value)
 #endif
 
+// Newer format_prompt_rerank() takes the decoded media of the query and of the
+// document. The gRPC rerank API is text only, so pass empty lists.
+#if LOCALAI_HAS_RERANK_FILES
+#define LOCALAI_RERANK_FILES_ARGS , std::vector<raw_buffer>(), std::vector<raw_buffer>()
+#else
+#define LOCALAI_RERANK_FILES_ARGS
+#endif
+
 // gRPC bearer token auth for distributed mode.
 // Reads LOCALAI_GRPC_AUTH_TOKEN from the environment. When set, rejects
 // requests without a matching "authorization: Bearer <token>" metadata header.
@@ -3114,7 +3122,7 @@ public:
 
             tasks.reserve(documents.size());
             for (size_t i = 0; i < documents.size(); i++) {
-                auto tmp = format_prompt_rerank(ctx_server.impl->model_tgt, ctx_server.impl->vocab, ctx_server.impl->mctx, request->query(), documents[i] LOCALAI_MTMD_INIT_OPT_ARG(ctx_server.impl->init_opt));
+                auto tmp = format_prompt_rerank(ctx_server.impl->model_tgt, ctx_server.impl->vocab, ctx_server.impl->mctx, request->query(), documents[i] LOCALAI_RERANK_FILES_ARGS LOCALAI_MTMD_INIT_OPT_ARG(ctx_server.impl->init_opt));
                 server_task task = server_task(SERVER_TASK_TYPE_RERANK);
                 task.id = rd.queue_tasks.get_new_id();
                 task.index = i;
diff --git a/backend/cpp/llama-cpp/prepare.sh b/backend/cpp/llama-cpp/prepare.sh
index 82d13b4fa..3d70f5447 100644
--- a/backend/cpp/llama-cpp/prepare.sh
+++ b/backend/cpp/llama-cpp/prepare.sh
@@ -134,6 +134,13 @@ if grep -q "common_chat_session & out_session" llama.cpp/tools/server/server-com
 else
     HAS_CHAT_SESSION=0
 fi
+# Multimodal rerank (query_files/doc_files on format_prompt_rerank,
+# ggml-org/llama.cpp#29625). Forks that branched earlier keep the old signature.
+if grep -q "query_files" llama.cpp/tools/server/server-common.h; then
+    HAS_RERANK_FILES=1
+else
+    HAS_RERANK_FILES=0
+fi
 cat > llama.cpp/tools/grpc-server/llama_compat.h <<EOF
 // Generated by backend/cpp/llama-cpp/prepare.sh. Do not edit.
 #pragma once
@@ -144,6 +151,7 @@ cat > llama.cpp/tools/grpc-server/llama_compat.h <<EOF
 #define LOCALAI_HAS_N_CPU_FFN_HELPER ${HAS_N_CPU_FFN_HELPER}
 #define LOCALAI_HAS_MOE_CACHE ${HAS_MOE_CACHE}
 #define LOCALAI_HAS_LAZY_MODE ${HAS_LAZY_MODE}
+#define LOCALAI_HAS_RERANK_FILES ${HAS_RERANK_FILES}
 EOF
 
 set +e