diff --git a/README.md b/README.md
index 56ff2c5..f752d97 100644
--- a/README.md
+++ b/README.md
@@ -14,6 +14,11 @@ if provider is not None:
     provider_models = list(provider.iter_models())
 ```
 
+Provider display names are not unique upstream, so `get_provider_by_name`
+returns `None` when a name is shared by more than one provider. Use
+`get_providers_by_name` to retrieve every match, or `get_provider_by_id` for an
+unambiguous lookup.
+
 ## Development
 
 Install dependencies and run the test suite with uv:
diff --git a/src/modelsdotdev/__init__.py b/src/modelsdotdev/__init__.py
index d1eccc0..fd6796d 100644
--- a/src/modelsdotdev/__init__.py
+++ b/src/modelsdotdev/__init__.py
@@ -16,6 +16,7 @@
     get_model_by_id,
     get_provider_by_id,
     get_provider_by_name,
+    get_providers_by_name,
     iter_models,
     iter_providers,
     parse_model_id,
@@ -37,6 +38,7 @@
     "get_model_by_id",
     "get_provider_by_id",
     "get_provider_by_name",
+    "get_providers_by_name",
     "iter_models",
     "iter_providers",
     "parse_model_id",
diff --git a/src/modelsdotdev/_internal/data.py b/src/modelsdotdev/_internal/data.py
index d0873bc..d2eb9d3 100644
--- a/src/modelsdotdev/_internal/data.py
+++ b/src/modelsdotdev/_internal/data.py
@@ -239,15 +239,31 @@ def qualified_id(self) -> str:
 DB_PATH = Path(__file__).parents[1] / "_db.sqlite"
 
 
-def get_provider_by_name(name: str) -> Provider | None:
-    """Return a provider by display name, using case-insensitive matching."""
+def get_providers_by_name(name: str) -> list[Provider]:
+    """Return all providers with the given display name, case-insensitively.
+
+    Upstream no longer guarantees provider names are unique, so a name may
+    resolve to more than one provider. Results are ordered by provider ID.
+    """
     with closing(_connect()) as connection:
-        row = connection.execute(
+        rows = connection.execute(
             f"SELECT {PROVIDER_COLUMNS} FROM providers "
-            "WHERE name = ? COLLATE NOCASE",
+            "WHERE name = ? COLLATE NOCASE ORDER BY id",
             (name,),
-        ).fetchone()
-        return None if row is None else _provider_from_row(row)
+        ).fetchall()
+        return [_provider_from_row(row) for row in rows]
+
+
+def get_provider_by_name(name: str) -> Provider | None:
+    """Return the sole provider with the given display name, else ``None``.
+
+    Matching is case-insensitive. When the name is ambiguous (shared by two or
+    more providers) ``None`` is returned, since there is no single correct
+    match; use :func:`get_providers_by_name` or look up by ID via
+    :func:`get_provider_by_id` instead.
+    """
+    providers = get_providers_by_name(name)
+    return providers[0] if len(providers) == 1 else None
 
 
 def get_provider_by_id(provider_id: str) -> Provider | None:
diff --git a/src/modelsdotdev/_internal/schema.py b/src/modelsdotdev/_internal/schema.py
index 48260e7..318465d 100644
--- a/src/modelsdotdev/_internal/schema.py
+++ b/src/modelsdotdev/_internal/schema.py
@@ -87,7 +87,7 @@ def _schema_sql(tables: tuple[Table, ...]) -> str:
         Column(name="env", definition="TEXT NOT NULL"),
     ),
     indexes=(
-        "CREATE UNIQUE INDEX providers_name_nocase_idx "
+        "CREATE INDEX providers_name_nocase_idx "
         "ON providers(name COLLATE NOCASE);",
     ),
 )
diff --git a/tests/test_api.py b/tests/test_api.py
index 14c3905..e7c0a03 100644
--- a/tests/test_api.py
+++ b/tests/test_api.py
@@ -1,4 +1,5 @@
 import sqlite3
+from collections import Counter
 from contextlib import closing
 
 import pytest
@@ -19,6 +20,7 @@
     get_model_by_id,
     get_provider_by_id,
     get_provider_by_name,
+    get_providers_by_name,
     iter_models,
     iter_providers,
     parse_model_id,
@@ -34,17 +36,41 @@ def test_provider_iteration_and_lookup_use_real_database() -> None:
         key=lambda provider: provider.name.lower(),
     )
     assert len({provider.id for provider in providers}) == len(providers)
-    assert len({provider.name.lower() for provider in providers}) == len(
-        providers,
+
+    # Upstream no longer guarantees provider names are unique.
+    name_counts = Counter(provider.name.lower() for provider in providers)
+
+    # A provider whose name is unique (case-insensitively) resolves by name.
+    unique = next(
+        provider
+        for provider in providers
+        if name_counts[provider.name.lower()] == 1
+    )
+    assert isinstance(unique, Provider)
+    assert get_provider_by_id(unique.id) == unique
+    assert get_provider_by_name(unique.name) == unique
+    assert get_provider_by_name(unique.name.upper()) == unique
+    assert get_providers_by_name(unique.name) == [unique]
+
+    # An ambiguous name (shared by 2+ providers) resolves to None via the
+    # singular getter but is fully enumerated by the plural getter.
+    ambiguous = next(
+        (
+            provider.name
+            for provider in providers
+            if name_counts[provider.name.lower()] > 1
+        ),
+        None,
     )
+    if ambiguous is not None:
+        matches = get_providers_by_name(ambiguous)
+        assert len(matches) == name_counts[ambiguous.lower()]
+        assert matches == get_providers_by_name(ambiguous.upper())
+        assert get_provider_by_name(ambiguous) is None
 
-    provider = providers[0]
-    assert isinstance(provider, Provider)
-    assert get_provider_by_id(provider.id) == provider
-    assert get_provider_by_name(provider.name) == provider
-    assert get_provider_by_name(provider.name.upper()) == provider
     assert get_provider_by_id("missing-provider") is None
     assert get_provider_by_name("missing provider") is None
+    assert get_providers_by_name("missing provider") == []
 
 
 def test_model_iteration_and_lookup_use_real_database() -> None: