feat(registry): resolve registry-only gateways without an enum member

scripts/generate_api_docs.py

@@ -434,9 +434,15 @@ def generate_any_llm_page() -> str:
 | `model` | `str` | Combined identifier in `"provider:model"` format (e.g., `"openai:gpt-4.1-mini"`). The legacy `"provider/model"` format is also accepted but deprecated. |"""
     )
     parts.append("")
-    parts.append("**Returns:** A `(LLMProvider, model_name)` tuple.")
+    parts.append(
+        "**Returns:** A `(provider, model_name)` tuple. `provider` is an `LLMProvider` member "
+        "when the provider has one, and the bare name for a config-only registry gateway."
+    )
     parts.append("")
-    parts.append("**Raises:** `ValueError` if the string does not contain a `:` or `/` delimiter.")
+    parts.append(
+        "**Raises:** `ValueError` if the string does not contain a `:` or `/` delimiter, or "
+        "`UnsupportedProviderError` if the provider is not resolvable."
+    )
     parts.append("")
     parts.append(
         """```python

src/any_llm/any_llm.py

@@ -250,7 +250,9 @@ def _create_provider(
         if registry_class is not None:
             return registry_class(api_key=api_key, api_base=api_base, **kwargs)
 
-        provider_key = LLMProvider.from_string(provider_key).value
+        # Resolve through the shared resolver so the unsupported-provider error lists
+        # registry gateways too, not just the enum.
+        provider_key = str(cls.resolve_provider_key(provider_key))
 
         provider_class_name = f"{provider_key.capitalize()}Provider"
         provider_module_name = f"{provider_key}"
@@ -296,7 +298,9 @@ def get_provider_class(cls, provider_key: str | LLMProvider) -> type[AnyLLM]:
         if registry_class is not None:
             return registry_class
 
-        provider_key = LLMProvider.from_string(provider_key).value
+        # Resolve through the shared resolver so the unsupported-provider error lists
+        # registry gateways too, not just the enum.
+        provider_key = str(cls.resolve_provider_key(provider_key))
 
         provider_class_name = f"{provider_key.capitalize()}Provider"
         provider_module_name = f"{provider_key}"
@@ -312,10 +316,54 @@ def get_provider_class(cls, provider_key: str | LLMProvider) -> type[AnyLLM]:
         provider_class: type[AnyLLM] = getattr(module, provider_class_name)
         return provider_class
 
+    @staticmethod
+    def get_registry_provider_names() -> list[str]:
+        """Names of config-only gateways that exist only as registry rows.
+
+        Registry rows do not need an ``LLMProvider`` member, so these names are
+        absent from the enum and have to be added to any enumeration of
+        providers explicitly.
+        """
+        from any_llm.providers.registry import PROVIDER_REGISTRY
+
+        enum_values = {provider.value for provider in LLMProvider}
+        return sorted(name for name in PROVIDER_REGISTRY if name not in enum_values)
+
     @classmethod
     def get_supported_providers(cls) -> list[str]:
-        """Get a list of supported provider keys."""
-        return [provider.value for provider in LLMProvider]
+        """Get a list of supported provider keys.
+
+        Includes registry-only gateways, which resolve by name without an
+        ``LLMProvider`` member.
+        """
+        return [provider.value for provider in LLMProvider] + cls.get_registry_provider_names()
+
+    @classmethod
+    def resolve_provider_key(cls, provider_key: str | LLMProvider) -> str | LLMProvider:
+        """Resolve a provider key to an ``LLMProvider`` member where one exists.
+
+        Registry-only gateways have no enum member, so their name is returned
+        unchanged. Everything downstream (``create``, ``get_provider_class``)
+        accepts either form.
+
+        Raises:
+            UnsupportedProviderError: The key is neither an enum member nor a
+                registry row.
+
+        """
+        if isinstance(provider_key, LLMProvider):
+            return provider_key
+        # Match LLMProvider.from_string's normalization so both resolution paths
+        # accept the same spellings.
+        normalized = provider_key.strip().lower()
+        try:
+            return LLMProvider(normalized)
+        except ValueError:
+            from any_llm.providers.registry import get_registry_config
+
+            if get_registry_config(normalized) is not None:
+                return normalized
+            raise UnsupportedProviderError(provider_key, cls.get_supported_providers()) from None
 
     @classmethod
     def get_all_provider_metadata(cls) -> list[ProviderMetadata]:
@@ -337,21 +385,29 @@ def get_all_provider_metadata(cls) -> list[ProviderMetadata]:
 
     @classmethod
     def get_provider_enum(cls, provider_key: str) -> LLMProvider:
-        """Convert a string provider key to a ProviderName enum."""
+        """Convert a string provider key to a ProviderName enum.
+
+        Registry-only gateways have no enum member, so this raises for them even
+        though they are resolvable. Use ``resolve_provider_key`` to accept both.
+        """
         try:
             return LLMProvider(provider_key)
         except ValueError as e:
-            supported = [provider.value for provider in LLMProvider]
-            raise UnsupportedProviderError(provider_key, supported) from e
+            # Report everything resolvable, not just the enum, so the message does
+            # not omit registry-only gateways.
+            raise UnsupportedProviderError(provider_key, cls.get_supported_providers()) from e
 
     @classmethod
-    def split_model_provider(cls, model: str) -> tuple[LLMProvider, str]:
+    def split_model_provider(cls, model: str) -> tuple[str | LLMProvider, str]:
         """Extract the provider key from the model identifier.
 
         Supports both new format 'provider:model' (e.g., 'mistral:mistral-small')
         and legacy format 'provider/model' (e.g., 'mistral/mistral-small').
 
         The legacy format will be deprecated in version 1.0.
+
+        Returns an ``LLMProvider`` member when the provider has one, and the bare
+        name for registry-only gateways. Both forms are accepted by ``create``.
         """
         colon_index = model.find(":")
         slash_index = model.find("/")
@@ -376,7 +432,7 @@ def split_model_provider(cls, model: str) -> tuple[LLMProvider, str]:
         if not provider or not model_name:
             msg = f"Invalid model format. Expected 'provider:model' or 'provider/model', got '{model}'"
             raise ValueError(msg)
-        return cls.get_provider_enum(provider), model_name
+        return cls.resolve_provider_key(provider), model_name
 
     @staticmethod
     @abstractmethod

src/any_llm/api.py

@@ -95,7 +95,7 @@ def completion(
     if provider is None:
         provider_key, model_id = AnyLLM.split_model_provider(model)
     else:
-        provider_key = LLMProvider.from_string(provider)
+        provider_key = AnyLLM.resolve_provider_key(provider)
         model_id = model
 
     llm = AnyLLM.create(
@@ -204,7 +204,7 @@ async def acompletion(
     if provider is None:
         provider_key, model_id = AnyLLM.split_model_provider(model)
     else:
-        provider_key = LLMProvider.from_string(provider)
+        provider_key = AnyLLM.resolve_provider_key(provider)
         model_id = model
 
     llm = AnyLLM.create(
@@ -345,7 +345,7 @@ def responses(
     if provider is None:
         provider_key, model_id = AnyLLM.split_model_provider(model)
     else:
-        provider_key = LLMProvider.from_string(provider)
+        provider_key = AnyLLM.resolve_provider_key(provider)
         model_id = model
 
     llm = AnyLLM.create(
@@ -494,7 +494,7 @@ async def aresponses(
     if provider is None:
         provider_key, model_id = AnyLLM.split_model_provider(model)
     else:
-        provider_key = LLMProvider.from_string(provider)
+        provider_key = AnyLLM.resolve_provider_key(provider)
         model_id = model
 
     llm = AnyLLM.create(
@@ -598,7 +598,7 @@ def messages(
     if provider is None:
         provider_key, model_id = AnyLLM.split_model_provider(model)
     else:
-        provider_key = LLMProvider.from_string(provider)
+        provider_key = AnyLLM.resolve_provider_key(provider)
         model_id = model
 
     llm = AnyLLM.create(provider_key, api_key=api_key, api_base=api_base, **client_args or {})
@@ -682,7 +682,7 @@ async def amessages(
     if provider is None:
         provider_key, model_id = AnyLLM.split_model_provider(model)
     else:
-        provider_key = LLMProvider.from_string(provider)
+        provider_key = AnyLLM.resolve_provider_key(provider)
         model_id = model
 
     llm = AnyLLM.create(provider_key, api_key=api_key, api_base=api_base, **client_args or {})
@@ -737,7 +737,7 @@ def embedding(
     if provider is None:
         provider_key, model_name = AnyLLM.split_model_provider(model)
     else:
-        provider_key = LLMProvider.from_string(provider)
+        provider_key = AnyLLM.resolve_provider_key(provider)
         model_name = model
 
     llm = AnyLLM.create(provider_key, api_key=api_key, api_base=api_base, **client_args or {})
@@ -775,7 +775,7 @@ async def aembedding(
     if provider is None:
         provider_key, model_name = AnyLLM.split_model_provider(model)
     else:
-        provider_key = LLMProvider.from_string(provider)
+        provider_key = AnyLLM.resolve_provider_key(provider)
         model_name = model
 
     llm = AnyLLM.create(provider_key, api_key=api_key, api_base=api_base, **client_args or {})
@@ -823,7 +823,7 @@ def image_generation(
     if provider is None:
         provider_key, model_name = AnyLLM.split_model_provider(model)
     else:
-        provider_key = LLMProvider.from_string(provider)
+        provider_key = AnyLLM.resolve_provider_key(provider)
         model_name = model
 
     llm = AnyLLM.create(provider_key, api_key=api_key, api_base=api_base, **client_args or {})
@@ -881,7 +881,7 @@ async def aimage_generation(
     if provider is None:
         provider_key, model_name = AnyLLM.split_model_provider(model)
     else:
-        provider_key = LLMProvider.from_string(provider)
+        provider_key = AnyLLM.resolve_provider_key(provider)
         model_name = model
 
     llm = AnyLLM.create(provider_key, api_key=api_key, api_base=api_base, **client_args or {})
@@ -942,7 +942,7 @@ def transcription(
     if provider is None:
         provider_key, model_name = AnyLLM.split_model_provider(model)
     else:
-        provider_key = LLMProvider.from_string(provider)
+        provider_key = AnyLLM.resolve_provider_key(provider)
         model_name = model
 
     llm = AnyLLM.create(provider_key, api_key=api_key, api_base=api_base, **client_args or {})
@@ -1000,7 +1000,7 @@ async def atranscription(
     if provider is None:
         provider_key, model_name = AnyLLM.split_model_provider(model)
     else:
-        provider_key = LLMProvider.from_string(provider)
+        provider_key = AnyLLM.resolve_provider_key(provider)
         model_name = model
 
     llm = AnyLLM.create(provider_key, api_key=api_key, api_base=api_base, **client_args or {})
@@ -1056,7 +1056,7 @@ def speech(
     if provider is None:
         provider_key, model_name = AnyLLM.split_model_provider(model)
     else:
-        provider_key = LLMProvider.from_string(provider)
+        provider_key = AnyLLM.resolve_provider_key(provider)
         model_name = model
 
     llm = AnyLLM.create(provider_key, api_key=api_key, api_base=api_base, **client_args or {})
@@ -1109,7 +1109,7 @@ async def aspeech(
     if provider is None:
         provider_key, model_name = AnyLLM.split_model_provider(model)
     else:
-        provider_key = LLMProvider.from_string(provider)
+        provider_key = AnyLLM.resolve_provider_key(provider)
         model_name = model
 
     llm = AnyLLM.create(provider_key, api_key=api_key, api_base=api_base, **client_args or {})
@@ -1163,7 +1163,7 @@ def moderation(
     if provider is None:
         provider_key, model_name = AnyLLM.split_model_provider(model)
     else:
-        provider_key = LLMProvider.from_string(provider)
+        provider_key = AnyLLM.resolve_provider_key(provider)
         model_name = model
 
     llm = AnyLLM.create(provider_key, api_key=api_key, api_base=api_base, **client_args or {})
@@ -1184,7 +1184,7 @@ async def amoderation(
     if provider is None:
         provider_key, model_name = AnyLLM.split_model_provider(model)
     else:
-        provider_key = LLMProvider.from_string(provider)
+        provider_key = AnyLLM.resolve_provider_key(provider)
         model_name = model
 
     llm = AnyLLM.create(provider_key, api_key=api_key, api_base=api_base, **client_args or {})
@@ -1204,7 +1204,7 @@ def _resolve_rerank_target(
     if provider is None:
         provider_key, model_name = AnyLLM.split_model_provider(model)
     else:
-        provider_key = LLMProvider.from_string(provider)
+        provider_key = AnyLLM.resolve_provider_key(provider)
         model_name = model
 
     if top_n is not None:
@@ -1287,7 +1287,7 @@ def list_models(
     **kwargs: Any,
 ) -> Sequence[Model]:
     """List available models for a provider."""
-    llm = AnyLLM.create(LLMProvider.from_string(provider), api_key=api_key, api_base=api_base, **client_args or {})
+    llm = AnyLLM.create(AnyLLM.resolve_provider_key(provider), api_key=api_key, api_base=api_base, **client_args or {})
     return llm.list_models(**kwargs)
 
 
@@ -1299,7 +1299,7 @@ async def alist_models(
     **kwargs: Any,
 ) -> Sequence[Model]:
     """List available models for a provider asynchronously."""
-    llm = AnyLLM.create(LLMProvider.from_string(provider), api_key=api_key, api_base=api_base, **client_args or {})
+    llm = AnyLLM.create(AnyLLM.resolve_provider_key(provider), api_key=api_key, api_base=api_base, **client_args or {})
     return await llm.alist_models(**kwargs)
 
 
@@ -1332,7 +1332,7 @@ def create_batch(
         The created batch object
 
     """
-    llm = AnyLLM.create(LLMProvider.from_string(provider), api_key=api_key, api_base=api_base, **client_args or {})
+    llm = AnyLLM.create(AnyLLM.resolve_provider_key(provider), api_key=api_key, api_base=api_base, **client_args or {})
     return llm.create_batch(
         input_file_path=input_file_path,
         endpoint=endpoint,
@@ -1371,7 +1371,7 @@ async def acreate_batch(
         The created batch object
 
     """
-    llm = AnyLLM.create(LLMProvider.from_string(provider), api_key=api_key, api_base=api_base, **client_args or {})
+    llm = AnyLLM.create(AnyLLM.resolve_provider_key(provider), api_key=api_key, api_base=api_base, **client_args or {})
     return await llm.acreate_batch(
         input_file_path=input_file_path,
         endpoint=endpoint,
@@ -1404,7 +1404,7 @@ def retrieve_batch(
         The batch object
 
     """
-    llm = AnyLLM.create(LLMProvider.from_string(provider), api_key=api_key, api_base=api_base, **client_args or {})
+    llm = AnyLLM.create(AnyLLM.resolve_provider_key(provider), api_key=api_key, api_base=api_base, **client_args or {})
     return llm.retrieve_batch(batch_id, **kwargs)
 
 
@@ -1431,7 +1431,7 @@ async def aretrieve_batch(
         The batch object
 
     """
-    llm = AnyLLM.create(LLMProvider.from_string(provider), api_key=api_key, api_base=api_base, **client_args or {})
+    llm = AnyLLM.create(AnyLLM.resolve_provider_key(provider), api_key=api_key, api_base=api_base, **client_args or {})
     return await llm.aretrieve_batch(batch_id, **kwargs)
 
 
@@ -1458,7 +1458,7 @@ def cancel_batch(
         The cancelled batch object
 
     """
-    llm = AnyLLM.create(LLMProvider.from_string(provider), api_key=api_key, api_base=api_base, **client_args or {})
+    llm = AnyLLM.create(AnyLLM.resolve_provider_key(provider), api_key=api_key, api_base=api_base, **client_args or {})
     return llm.cancel_batch(batch_id, **kwargs)
 
 
@@ -1485,7 +1485,7 @@ async def acancel_batch(
         The cancelled batch object
 
     """
-    llm = AnyLLM.create(LLMProvider.from_string(provider), api_key=api_key, api_base=api_base, **client_args or {})
+    llm = AnyLLM.create(AnyLLM.resolve_provider_key(provider), api_key=api_key, api_base=api_base, **client_args or {})
     return await llm.acancel_batch(batch_id, **kwargs)
 
 
@@ -1514,7 +1514,7 @@ def list_batches(
         A list of batch objects
 
     """
-    llm = AnyLLM.create(LLMProvider.from_string(provider), api_key=api_key, api_base=api_base, **client_args or {})
+    llm = AnyLLM.create(AnyLLM.resolve_provider_key(provider), api_key=api_key, api_base=api_base, **client_args or {})
     return llm.list_batches(after=after, limit=limit, **kwargs)
 
 
@@ -1543,7 +1543,7 @@ async def alist_batches(
         A list of batch objects
 
     """
-    llm = AnyLLM.create(LLMProvider.from_string(provider), api_key=api_key, api_base=api_base, **client_args or {})
+    llm = AnyLLM.create(AnyLLM.resolve_provider_key(provider), api_key=api_key, api_base=api_base, **client_args or {})
     return await llm.alist_batches(after=after, limit=limit, **kwargs)
 
 
@@ -1570,7 +1570,7 @@ def retrieve_batch_results(
         The batch results containing per-request outcomes.
 
     """
-    llm = AnyLLM.create(LLMProvider.from_string(provider), api_key=api_key, api_base=api_base, **client_args or {})
+    llm = AnyLLM.create(AnyLLM.resolve_provider_key(provider), api_key=api_key, api_base=api_base, **client_args or {})
     return llm.retrieve_batch_results(batch_id, **kwargs)
 
 
@@ -1597,5 +1597,5 @@ async def aretrieve_batch_results(
         The batch results containing per-request outcomes.
 
     """
-    llm = AnyLLM.create(LLMProvider.from_string(provider), api_key=api_key, api_base=api_base, **client_args or {})
+    llm = AnyLLM.create(AnyLLM.resolve_provider_key(provider), api_key=api_key, api_base=api_base, **client_args or {})
     return await llm.aretrieve_batch_results(batch_id, **kwargs)