[Bugfix] Make Gemma3 MM V0 only for now (vllm-project#14971)

Signed-off-by: Roger Wang <[email protected]>

Author: ywang96

docs/source/models/supported_models.md

@@ -763,7 +763,7 @@ See [this page](#generative-models) for more information on how to use generativ
   * `google/gemma-3-4b-it`, `google/gemma-3-27b-it`, etc.
   * ✅︎
   * ✅︎
-  * ⚠️
+  *
 - * `GLM4VForCausalLM`<sup>^</sup>
   * GLM-4V
   * T + I
@@ -948,8 +948,11 @@ V1 currently uses a simplified attention pattern:
 - Uses causal attention for all tokens, including image tokens
 - Generates reasonable outputs but does not match the original model's attention for text + image inputs
 - Will be updated in the future to support the correct behavior
+- Does not support `"do_pan_and_scan": True`
 
 This limitation exists because the model's mixed attention pattern (bidirectional for images, causal otherwise) is not yet supported by vLLM's attention backends.
+
+For these reasons, `Gemma3ForConditionalGeneration` is supported only on V0 at the moment.
 :::
 
 :::{note}

vllm/model_executor/models/gemma3_mm.py

@@ -25,7 +25,7 @@
 from vllm.sequence import IntermediateTensors
 
 from .interfaces import (MultiModalEmbeddings, SupportsLoRA,
-                         SupportsMultiModal, SupportsPP)
+                         SupportsMultiModal, SupportsPP, SupportsV0Only)
 from .siglip import SiglipVisionModel
 from .utils import (AutoWeightsLoader, flatten_bn, init_vllm_registered_model,
                     maybe_prefix, merge_multimodal_embeddings)
@@ -374,7 +374,7 @@ def forward(self, vision_outputs: torch.Tensor):
                                         info=Gemma3ProcessingInfo,
                                         dummy_inputs=Gemma3DummyInputsBuilder)
 class Gemma3ForConditionalGeneration(nn.Module, SupportsMultiModal, SupportsPP,
-                                     SupportsLoRA):
+                                     SupportsLoRA, SupportsV0Only):
     packed_modules_mapping = {
         "qkv_proj": [
             "q_proj",