diff --git a/gpt4all-chat/chatllm.cpp b/gpt4all-chat/chatllm.cpp
index 3b6e3c6c..153f5da0 100644
--- a/gpt4all-chat/chatllm.cpp
+++ b/gpt4all-chat/chatllm.cpp
@@ -37,6 +37,7 @@ using namespace Qt::Literals::StringLiterals;
 //#define DEBUG
 //#define DEBUG_MODEL_LOADING
 
+#define GPTJ_INTERNAL_STATE_VERSION  0 // GPT-J is gone but old chats still use this
 #define LLAMA_INTERNAL_STATE_VERSION 0
 
 class LLModelStore {
@@ -1055,6 +1056,7 @@ bool ChatLLM::serialize(QDataStream &stream, int version, bool serializeKV)
     if (version > 1) {
         stream << m_llModelType;
         switch (m_llModelType) {
+        case GPTJ_: stream << GPTJ_INTERNAL_STATE_VERSION; break;
         case LLAMA_: stream << LLAMA_INTERNAL_STATE_VERSION; break;
         default: Q_UNREACHABLE();
         }
diff --git a/gpt4all-chat/chatllm.h b/gpt4all-chat/chatllm.h
index fd3a7a46..252b77bd 100644
--- a/gpt4all-chat/chatllm.h
+++ b/gpt4all-chat/chatllm.h
@@ -28,9 +28,12 @@ using namespace Qt::Literals::StringLiterals;
 
 class QDataStream;
 
+// NOTE: values serialized to disk, do not change or reuse
 enum LLModelType {
-    LLAMA_,
-    API_,
+    GPTJ_  = 0, // no longer used
+    LLAMA_ = 1,
+    API_   = 2,
+    BERT_  = 3, // no longer used
 };
 
 class ChatLLM;