test(embeddings): add parametrized tests for dimensions drop_params behaviour

Cover six scenarios for openai_compatible_providers:
- non-text-embedding-3 model + drop_params=True  → dimensions dropped
- non-text-embedding-3 model + drop_params=False → dimensions forwarded
- text-embedding-3-* model (any drop_params)     → dimensions always kept

Regression test for #23119.

tests/test_litellm/test_utils.py

@@ -474,6 +474,52 @@ def test_cohere_embedding_optional_params():
     assert optional_params is not None
 
 
+@pytest.mark.parametrize(
+    "model,custom_llm_provider,drop_params,expect_dimensions",
+    [
+        # hosted_vllm with a non-text-embedding-3 model and drop_params=True → drop
+        ("intfloat/e5-large-v2", "hosted_vllm", True, False),
+        # hosted_vllm with a non-text-embedding-3 model and drop_params=False → keep
+        ("intfloat/e5-large-v2", "hosted_vllm", False, True),
+        # text-embedding-3 model should always keep dimensions
+        ("text-embedding-3-small", "openai_compatible_providers", True, True),
+        ("text-embedding-3-small", "openai_compatible_providers", False, True),
+        # openrouter with a non-text-embedding-3 model and drop_params=True → drop
+        ("jina-embeddings-v3", "openai_compatible_providers", True, False),
+        # openrouter without drop_params → keep (don't block providers that support dimensions)
+        ("jina-embeddings-v3", "openai_compatible_providers", False, True),
+    ],
+)
+def test_openai_compatible_embedding_dimensions_drop_params(
+    model, custom_llm_provider, drop_params, expect_dimensions
+):
+    """dimensions should only be dropped for non-text-embedding-3 models when drop_params=True.
+
+    Regression test for https://github.com/BerriAI/litellm/issues/23119 — hosted_vllm
+    endpoints return 422 when `dimensions` is forwarded, but other compatible providers
+    (Jina AI, Cohere via OpenRouter) legitimately support the parameter.
+    """
+    from litellm import get_optional_params_embeddings
+
+    result = get_optional_params_embeddings(
+        model=model,
+        custom_llm_provider=custom_llm_provider,
+        input="hello",
+        dimensions=512,
+        drop_params=drop_params,
+    )
+    if expect_dimensions:
+        assert "dimensions" in result, (
+            f"Expected 'dimensions' to be present for model={model!r}, "
+            f"drop_params={drop_params}"
+        )
+    else:
+        assert "dimensions" not in result, (
+            f"Expected 'dimensions' to be dropped for model={model!r}, "
+            f"drop_params={drop_params}"
+        )
+
+
 def validate_model_cost_values(model_data, exceptions=None):
     """
     Validates that cost values in model data do not exceed 1.