diff --git a/optimum/exporters/openvino/model_configs.py b/optimum/exporters/openvino/model_configs.py
index b8ffcdcf26..a7c3462f4d 100644
--- a/optimum/exporters/openvino/model_configs.py
+++ b/optimum/exporters/openvino/model_configs.py
@@ -3659,9 +3659,6 @@ class M2M100OpenVINOConfig(BartOpenVINOConfig):
     pass
 
 
-@register_in_tasks_manager(
-    "deepseek_v3", *["text-generation", "text-generation-with-past"], library_name="transformers"
-)
 @register_in_tasks_manager(
     "deepseek_v2", *["text-generation", "text-generation-with-past"], library_name="transformers"
 )
@@ -3671,6 +3668,15 @@ class DeepseekOpenVINOConfig(MiniCPM3OpenVINOConfig):
     _MODEL_PATCHER = DeepseekPatcher
 
 
+@register_in_tasks_manager(
+    "deepseek_v3", *["text-generation", "text-generation-with-past"], library_name="transformers"
+)
+class DeepseekVOpenVINOConfig(LlamaOpenVINOConfig):
+    MIN_TRANSFORMERS_VERSION = "4.51.0"
+    DUMMY_INPUT_GENERATOR_CLASSES = (DummyTextInputGenerator, OVMiniCPM3DummyPastKeyValuesGenerator)
+    DUMMY_PKV_GENERATOR_CLASS = OVMiniCPM3DummyPastKeyValuesGenerator
+
+
 @register_in_tasks_manager("got_ocr2", *["image-to-text", "image-text-to-text"], library_name="transformers")
 class GotOCR2OpenVINOConfig(BaseVLMOpenVINOConfig):
     MIN_TRANSFORMERS_VERSION = "4.49.0"
@@ -4251,6 +4257,14 @@ def generate_dummy_inputs(self, framework: str = "pt", **kwargs):
         return dummy_inputs
 
 
+@register_in_tasks_manager("ernie4_5", *["text-generation", "text-generation-with-past"], library_name="transformers")
+class ErnieOpenVINOConfig(LlamaOpenVINOConfig):
+    MIN_TRANSFORMERS_VERSION = "4.54.0"
+    DUMMY_INPUT_GENERATOR_CLASSES = (DummyTextInputGenerator, GemmaDummyPastKeyValuesGenerator)
+    DUMMY_PKV_GENERATOR_CLASS = GemmaDummyPastKeyValuesGenerator
+    NORMALIZED_CONFIG_CLASS = NormalizedTextConfig
+
+
 @register_in_tasks_manager(
     "gpt2",
     *[
diff --git a/tests/openvino/test_decoder.py b/tests/openvino/test_decoder.py
index 007b256a7e..2a3529e88e 100644
--- a/tests/openvino/test_decoder.py
+++ b/tests/openvino/test_decoder.py
@@ -93,9 +93,6 @@ class OVModelForCausalLMIntegrationTest(unittest.TestCase):
 
     if is_transformers_version(">=", "4.46.0"):
         SUPPORTED_ARCHITECTURES += ("glm", "mistral-nemo", "minicpm3", "phi3-moe")
-        # openvino 2025.0 required for disabling check_trace
-        if is_openvino_version(">=", "2025.0"):
-            SUPPORTED_ARCHITECTURES += ("deepseek",)
 
         # gptq and awq install disabled for windows test environment
         if platform.system() != "Windows":
@@ -109,7 +106,7 @@ class OVModelForCausalLMIntegrationTest(unittest.TestCase):
         SUPPORTED_ARCHITECTURES += ("gemma3_text",)
 
     if is_transformers_version(">=", "4.51.0"):
-        SUPPORTED_ARCHITECTURES += ("qwen3", "qwen3_moe")
+        SUPPORTED_ARCHITECTURES += ("qwen3", "qwen3_moe", "deepseek_v3")
 
     if is_transformers_version(">=", "4.51.3"):
         SUPPORTED_ARCHITECTURES += ("glm4",)
@@ -120,6 +117,7 @@ class OVModelForCausalLMIntegrationTest(unittest.TestCase):
     if is_transformers_version(">=", "4.54.0"):
         # remote code models differs after transformers v4.54
         SUPPORTED_ARCHITECTURES = tuple(set(SUPPORTED_ARCHITECTURES) - {"minicpm", "minicpm3", "arctic", "deepseek"})
+        SUPPORTED_ARCHITECTURES += ("ernie4_5",)
 
     if is_transformers_version(">=", "4.55.0"):
         SUPPORTED_ARCHITECTURES += ("gpt_oss", "gpt_oss_mxfp4")
@@ -144,7 +142,6 @@ class OVModelForCausalLMIntegrationTest(unittest.TestCase):
         "exaone",
         "decilm",
         "minicpm3",
-        "deepseek",
     )
 
     EXPECTED_NUM_SDPA = {
@@ -158,6 +155,8 @@ class OVModelForCausalLMIntegrationTest(unittest.TestCase):
         "chatglm": 2,
         "codegen": 5,
         "codegen2": 2,
+        "deepseek_v3": 2,
+        "ernie4_5": 4,
         "gpt2": 5,
         "gptj": 5,
         "gpt_neo": 4,
@@ -202,7 +201,6 @@ class OVModelForCausalLMIntegrationTest(unittest.TestCase):
         "mistral-nemo": 8,
         "minicpm3": 6,
         "phi3-moe": 2,
-        "deepseek": 2,
         "opt_gptq": 12,
         "mixtral_awq": 2,
         "gemma3_text": 2,
@@ -324,10 +322,6 @@ def test_compare_to_transformers(self, model_arch):
 
         ov_outputs = ov_model.generate(**tokens, generation_config=gen_config)
 
-        # TODO: add back once https://huggingface.co/katuni4ka/tiny-random-minicpm3/discussions/1 merged (for all models) as current mdoeling incompatible with transformers >= v4.49
-        if model_arch in {"deepseek"} and is_transformers_version(">=", "4.49"):
-            self.skipTest("Incompatible modeling code")
-
         additional_inputs = {}
         # gemma2 does not support dynamic cache, it is unfair to compare dynamic cache result vs hybrid cache,
         # align cache representation in torch model
@@ -569,10 +563,6 @@ def test_beam_search(self, model_arch):
         if model_arch in ["qwen", "chatglm", "chatglm4"]:
             return
 
-        # TODO: add back once https://huggingface.co/katuni4ka/tiny-random-minicpm3/discussions/1 merged (for all models) as current mdoeling incompatible with transformers >= v4.49
-        if model_arch in {"deepseek"} and is_transformers_version(">=", "4.49"):
-            self.skipTest("Incompatible modeling code")
-
         tokenizer = AutoTokenizer.from_pretrained(model_id, trust_remote_code=model_arch in self.REMOTE_CODE_MODELS)
         if model_arch == "persimmon":
             tokenizer.pad_token_id = tokenizer.bos_token_id
diff --git a/tests/openvino/utils_tests.py b/tests/openvino/utils_tests.py
index f7fa0ca6bb..64acd41f78 100644
--- a/tests/openvino/utils_tests.py
+++ b/tests/openvino/utils_tests.py
@@ -66,7 +66,7 @@
     "deberta": "optimum-intel-internal-testing/tiny-random-deberta",
     "deberta-v2": "optimum-intel-internal-testing/tiny-random-DebertaV2Model",
     "decilm": "optimum-intel-internal-testing/tiny-random-decilm",
-    "deepseek": "optimum-intel-internal-testing/tiny-random-deepseek-v3",
+    "deepseek_v3": "hf-internal-testing/tiny-random-DeepseekV3ForCausalLM",
     "deit": "optimum-intel-internal-testing/tiny-random-DeiTModel",
     "convnext": "optimum-intel-internal-testing/tiny-random-convnext",
     "convnextv2": "optimum-intel-internal-testing/tiny-random-ConvNextV2Model",
@@ -75,6 +75,7 @@
     "donut-swin": "optimum-intel-internal-testing/tiny-random-DonutSwinModel",
     "detr": "optimum-intel-internal-testing/tiny-random-DetrModel",
     "electra": "optimum-intel-internal-testing/tiny-random-electra",
+    "ernie4_5": "optimum-internal-testing/tiny-random-Ernie4_5ForCausalLM",
     "esm": "optimum-intel-internal-testing/tiny-random-EsmModel",
     "exaone": "optimum-intel-internal-testing/tiny-random-exaone",
     "gemma": "optimum-intel-internal-testing/tiny-random-GemmaForCausalLM",