diff --git a/vllm/config/speculative.py b/vllm/config/speculative.py
index e2904de9aa72..0f8adc6e4ac2 100644
--- a/vllm/config/speculative.py
+++ b/vllm/config/speculative.py
@@ -1003,6 +1003,7 @@ def _verify_args(self) -> Self:
             "kimi_k25",
             "minimax_m2",
             "gemma4",
+            "laguna",
         ]
         if (
             self.method in ("eagle3", "extract_hidden_states", "dflash")
diff --git a/vllm/model_executor/models/laguna.py b/vllm/model_executor/models/laguna.py
index 08f35d691817..5bf5b7cb021e 100644
--- a/vllm/model_executor/models/laguna.py
+++ b/vllm/model_executor/models/laguna.py
@@ -39,7 +39,12 @@
     default_weight_loader,
     maybe_remap_kv_scale_name,
 )
-from vllm.model_executor.models.interfaces import SupportsLoRA, SupportsPP
+from vllm.model_executor.models.interfaces import (
+    EagleModelMixin,
+    SupportsEagle3,
+    SupportsLoRA,
+    SupportsPP,
+)
 from vllm.model_executor.models.utils import (
     AutoWeightsLoader,
     PPMissingLayer,
@@ -554,7 +559,7 @@ def forward(
 
 
 @support_torch_compile
-class LagunaModel(nn.Module):
+class LagunaModel(nn.Module, EagleModelMixin):
     def __init__(self, *, vllm_config: VllmConfig, prefix: str = ""):
         super().__init__()
 
@@ -633,8 +638,17 @@ def forward(
             hidden_states = intermediate_tensors["hidden_states"]
             residual = intermediate_tensors["residual"]
 
-        for layer in islice(self.layers, self.start_layer, self.end_layer):
+        aux_hidden_states = self._maybe_add_hidden_state(
+            [], self.start_layer, hidden_states, residual
+        )
+        for layer_idx, layer in enumerate(
+            islice(self.layers, self.start_layer, self.end_layer),
+            start=self.start_layer,
+        ):
             hidden_states, residual = layer(positions, hidden_states, residual)
+            self._maybe_add_hidden_state(
+                aux_hidden_states, layer_idx + 1, hidden_states, residual
+            )
 
         if not get_pp_group().is_last_rank:
             return IntermediateTensors(
@@ -642,6 +656,8 @@ def forward(
             )
 
         hidden_states, _ = self.norm(hidden_states, residual)
+        if len(aux_hidden_states) > 0:
+            return hidden_states, aux_hidden_states
         return hidden_states
 
     def get_expert_mapping(self) -> list[tuple[str, str, int, str]]:
@@ -821,7 +837,7 @@ def load_weights(self, weights: Iterable[tuple[str, torch.Tensor]]) -> set[str]:
         return loaded_params
 
 
-class LagunaForCausalLM(nn.Module, SupportsPP, SupportsLoRA):
+class LagunaForCausalLM(nn.Module, SupportsPP, SupportsLoRA, SupportsEagle3):
     fall_back_to_pt_during_load = False
 
     packed_modules_mapping = {