diff --git a/src/llama-context.cpp b/src/llama-context.cpp
--- a/src/llama-context.cpp
+++ b/src/llama-context.cpp
@@ -2133,7 +2133,7 @@ uint32_t llama_context::output_reserve(int32_t n_outputs) {
     const auto n_embd     = hparams.n_embd;
     const auto n_embd_out = hparams.n_embd_out();
 
-    bool has_logits     = true;
+    bool has_logits     = !(cparams.embeddings && !cparams.causal_attn);
     bool has_embd       = cparams.embeddings;
     bool has_embd_nextn = cparams.embeddings_nextn;