mirror of
https://github.com/BerriAI/litellm.git
synced 2026-11-17 09:34:48 +00:00
feat: list the Opper catalog from the gateway and add it to the dashboard
get_models reads the gateway's own model list so the 600+ model catalog tracks Opper rather than a static price table, with get_api_key and get_api_base resolving OPPER_* instead of inheriting OpenAI's. Registering the provider for model info is what makes that list reachable, so the simple half of get_provider_model_info became a frozen factory map to stay under the complexity ceiling. Opper is now selectable in the Add Model form with its own logo and placeholder.
This commit is contained in:
parent
220801e140
commit
c687f9ffcf
11 changed files with 344 additions and 154 deletions
|
|
@ -234,6 +234,7 @@ DEFAULT_REASONING_EFFORT_MINIMAL_THINKING_BUDGET: Final = int(
|
|||
|
||||
# Provider-specific API base URLs
|
||||
XAI_API_BASE: Final = "https://api.x.ai/v1"
|
||||
OPPER_API_BASE: Final = "https://api.opper.ai/v3/compat"
|
||||
OPEN_SANDBOX_API_BASE_ENV_VAR: Final = "OPEN_SANDBOX_API_BASE"
|
||||
OPEN_SANDBOX_API_KEY_ENV_VAR: Final = "OPEN_SANDBOX_API_KEY"
|
||||
OPEN_SANDBOX_DEFAULT_TEMPLATE: Final = "opensandbox/code-interpreter:v1.1.0"
|
||||
|
|
|
|||
|
|
@ -2,7 +2,11 @@ from typing import Final, cast
|
|||
from urllib.parse import urlparse
|
||||
|
||||
import litellm
|
||||
from litellm.constants import PROVIDERS_THAT_AUTHENTICATE_ON_PROVIDER_INFO, REPLICATE_MODEL_NAME_WITH_ID_LENGTH
|
||||
from litellm.constants import (
|
||||
OPPER_API_BASE,
|
||||
PROVIDERS_THAT_AUTHENTICATE_ON_PROVIDER_INFO,
|
||||
REPLICATE_MODEL_NAME_WITH_ID_LENGTH,
|
||||
)
|
||||
from litellm.litellm_core_utils.fallback_generalizations import (
|
||||
match_routing_generalization,
|
||||
)
|
||||
|
|
@ -699,8 +703,10 @@ def _get_openai_compatible_provider_info(
|
|||
|
||||
dynamic_api_key = api_key or get_secret_str("TENCENT_API_KEY")
|
||||
elif custom_llm_provider == "opper":
|
||||
api_base = api_base or get_secret("OPPER_API_BASE") or "https://api.opper.ai/v3/compat"
|
||||
# rebind-ok: provider chain assigns api_base per branch
|
||||
api_base = api_base or get_secret("OPPER_API_BASE") or OPPER_API_BASE
|
||||
|
||||
# rebind-ok: provider chain assigns dynamic_api_key per branch
|
||||
dynamic_api_key = api_key or get_secret_str("OPPER_API_KEY")
|
||||
elif custom_llm_provider == "fireworks_ai":
|
||||
# fireworks is openai compatible, we just need to set this to custom_openai and have the api_base be https://api.fireworks.ai/inference/v1
|
||||
|
|
|
|||
|
|
@ -7,8 +7,11 @@ from typing import Final
|
|||
|
||||
import httpx
|
||||
|
||||
import litellm
|
||||
from litellm._logging import verbose_logger
|
||||
from litellm.constants import OPPER_API_BASE
|
||||
from litellm.llms.base_llm.chat.transformation import BaseLLMException, LiteLLMLoggingObj
|
||||
from litellm.secret_managers.main import get_secret_str
|
||||
from litellm.types.llms.openai import AllMessageValues
|
||||
from litellm.types.utils import ModelResponse, ModelResponseStream, StreamingChoices
|
||||
|
||||
|
|
@ -47,7 +50,7 @@ class OpperConfig(OpenAIGPTConfig):
|
|||
) -> dict: # mutable-ok: mirrors the OpenAIGPTConfig signature
|
||||
request: Final = super().transform_request(model, messages, optional_params, litellm_params, headers)
|
||||
if request.get("stream"):
|
||||
stream_options: Final = dict(request.get("stream_options") or {})
|
||||
stream_options: Final = dict(request.get("stream_options") or {}) # mutable-ok: JSON request body
|
||||
stream_options.setdefault("include_usage", True)
|
||||
request["stream_options"] = stream_options
|
||||
return request
|
||||
|
|
@ -66,7 +69,7 @@ class OpperConfig(OpenAIGPTConfig):
|
|||
api_key: str | None = None,
|
||||
json_mode: bool | None = None,
|
||||
) -> ModelResponse:
|
||||
model_response = super().transform_response(
|
||||
transformed: Final = super().transform_response(
|
||||
model=model,
|
||||
raw_response=raw_response,
|
||||
model_response=model_response,
|
||||
|
|
@ -85,15 +88,15 @@ class OpperConfig(OpenAIGPTConfig):
|
|||
if response_json.get("usage"):
|
||||
response_cost: Final = response_json["usage"].get("cost")
|
||||
if response_cost is not None:
|
||||
if "additional_headers" not in model_response._hidden_params:
|
||||
model_response._hidden_params["additional_headers"] = {}
|
||||
model_response._hidden_params["additional_headers"]["llm_provider-x-litellm-response-cost"] = float(
|
||||
response_cost
|
||||
headers: Final = transformed._hidden_params.setdefault(
|
||||
"additional_headers",
|
||||
{}, # mutable-ok: _hidden_params carries plain dicts
|
||||
)
|
||||
headers["llm_provider-x-litellm-response-cost"] = float(response_cost)
|
||||
except (ValueError, TypeError, KeyError, AttributeError):
|
||||
verbose_logger.debug("Opper: could not extract usage.cost from response")
|
||||
|
||||
return model_response
|
||||
return transformed
|
||||
|
||||
def get_error_class(
|
||||
self,
|
||||
|
|
@ -107,6 +110,37 @@ class OpperConfig(OpenAIGPTConfig):
|
|||
headers=headers,
|
||||
)
|
||||
|
||||
@staticmethod
|
||||
def get_api_key(api_key: str | None = None) -> str | None:
|
||||
return api_key or get_secret_str("OPPER_API_KEY")
|
||||
|
||||
@staticmethod
|
||||
def get_api_base(api_base: str | None = None) -> str | None:
|
||||
return api_base or get_secret_str("OPPER_API_BASE") or OPPER_API_BASE
|
||||
|
||||
def get_models(
|
||||
self,
|
||||
api_key: str | None = None,
|
||||
api_base: str | None = None,
|
||||
) -> list[str]: # mutable-ok: mirrors the BaseLLMModelInfo signature
|
||||
resolved_api_base: Final = self.get_api_base(api_base) or OPPER_API_BASE
|
||||
resolved_api_key: Final = self.get_api_key(api_key)
|
||||
if resolved_api_key is None:
|
||||
raise ValueError("Opper API key is not set. Pass api_key or set OPPER_API_KEY")
|
||||
|
||||
response: Final = litellm.module_level_client.get(
|
||||
url=f"{resolved_api_base.rstrip('/')}/models",
|
||||
headers={"Authorization": f"Bearer {resolved_api_key}"}, # mutable-ok: httpx request headers
|
||||
)
|
||||
try:
|
||||
response.raise_for_status()
|
||||
except httpx.HTTPStatusError as e:
|
||||
raise ValueError(
|
||||
f"Failed to fetch models from Opper. Status code: {response.status_code}, Response: {response.text}"
|
||||
) from e
|
||||
|
||||
return [f"opper/{model['id']}" for model in response.json()["data"]] # mutable-ok: base returns list[str]
|
||||
|
||||
def get_model_response_iterator(
|
||||
self,
|
||||
streaming_response: Iterator[str] | AsyncIterator[str] | ModelResponse,
|
||||
|
|
@ -124,5 +158,5 @@ class OpperChatCompletionStreamingHandler(OpenAIChatCompletionStreamingHandler):
|
|||
def chunk_parser(self, chunk: dict) -> ModelResponseStream: # mutable-ok: mirrors the base chunk_parser signature
|
||||
parsed: Final = super().chunk_parser(chunk)
|
||||
if getattr(parsed, "usage", None) is not None and not parsed.choices:
|
||||
parsed.choices = [StreamingChoices()]
|
||||
parsed.choices = [StreamingChoices()] # mutable-ok: ModelResponseStream.choices is a list
|
||||
return parsed
|
||||
|
|
|
|||
|
|
@ -64,6 +64,7 @@ from litellm.constants import (
|
|||
AZURE_OPENAI_AUDIO_PROVIDERS,
|
||||
DEFAULT_MOCK_RESPONSE_COMPLETION_TOKEN_COUNT,
|
||||
DEFAULT_MOCK_RESPONSE_PROMPT_TOKEN_COUNT,
|
||||
OPPER_API_BASE,
|
||||
)
|
||||
from litellm.exceptions import LiteLLMUnknownProvider
|
||||
from litellm.integrations.custom_logger import CustomLogger
|
||||
|
|
@ -3478,9 +3479,7 @@ def _complete_opper(ctx: _CompletionDispatchContext) -> _CompletionDispatchResul
|
|||
stream: Final = ctx.stream
|
||||
timeout: Final = ctx.timeout
|
||||
|
||||
api_base: Final = (
|
||||
ctx.api_base or litellm.api_base or get_secret_str("OPPER_API_BASE") or "https://api.opper.ai/v3/compat"
|
||||
)
|
||||
api_base: Final = ctx.api_base or litellm.api_base or get_secret_str("OPPER_API_BASE") or OPPER_API_BASE
|
||||
api_key: Final = ctx.api_key or litellm.api_key or get_secret_str("OPPER_API_KEY")
|
||||
headers: Final = ctx.headers or litellm.headers
|
||||
|
||||
|
|
|
|||
|
|
@ -2505,6 +2505,34 @@
|
|||
],
|
||||
"default_model_placeholder": "gpt-3.5-turbo"
|
||||
},
|
||||
{
|
||||
"provider": "Opper",
|
||||
"provider_display_name": "Opper",
|
||||
"litellm_provider": "opper",
|
||||
"credential_fields": [
|
||||
{
|
||||
"key": "api_base",
|
||||
"label": "API Base",
|
||||
"placeholder": "https://api.opper.ai/v3/compat",
|
||||
"tooltip": null,
|
||||
"required": false,
|
||||
"field_type": "text",
|
||||
"options": null,
|
||||
"default_value": null
|
||||
},
|
||||
{
|
||||
"key": "api_key",
|
||||
"label": "API Key",
|
||||
"placeholder": null,
|
||||
"tooltip": null,
|
||||
"required": true,
|
||||
"field_type": "password",
|
||||
"options": null,
|
||||
"default_value": null
|
||||
}
|
||||
],
|
||||
"default_model_placeholder": "opper/anthropic/claude-haiku-4-5"
|
||||
},
|
||||
{
|
||||
"provider": "Oracle",
|
||||
"provider_display_name": "Oracle Cloud Infrastructure (OCI)",
|
||||
|
|
|
|||
|
|
@ -4700,11 +4700,13 @@ def get_optional_params(
|
|||
drop_params=bool(drop_params),
|
||||
)
|
||||
elif custom_llm_provider == "opper":
|
||||
optional_params = litellm.OpperConfig().map_openai_params( # rebind-ok: provider chain assigns optional_params per branch
|
||||
non_default_params=non_default_params,
|
||||
optional_params=optional_params,
|
||||
model=model,
|
||||
drop_params=(drop_params if drop_params is not None and isinstance(drop_params, bool) else False),
|
||||
optional_params = (
|
||||
litellm.OpperConfig().map_openai_params( # rebind-ok: provider chain assigns optional_params per branch
|
||||
non_default_params=non_default_params,
|
||||
optional_params=optional_params,
|
||||
model=model,
|
||||
drop_params=(drop_params if drop_params is not None and isinstance(drop_params, bool) else False),
|
||||
)
|
||||
)
|
||||
elif custom_llm_provider == "watsonx":
|
||||
optional_params = litellm.IBMWatsonXChatConfig().map_openai_params(
|
||||
|
|
@ -8107,6 +8109,23 @@ class ProviderConfigManager:
|
|||
# This is initialized lazily on first access to avoid circular imports
|
||||
_PROVIDER_CONFIG_MAP: dict[LlmProviders, tuple[Callable, bool]] | None = None
|
||||
|
||||
# Model-info providers whose class is already exported on the litellm module.
|
||||
# Providers that need a lazy import or the model name stay in get_provider_model_info.
|
||||
_MODEL_INFO_FACTORIES: Mapping[LlmProviders, Callable[[], BaseLLMModelInfo]] = MappingProxyType(
|
||||
{
|
||||
LlmProviders.FIREWORKS_AI: lambda: litellm.FireworksAIConfig(),
|
||||
LlmProviders.OPENAI: lambda: litellm.OpenAIGPTConfig(),
|
||||
LlmProviders.GEMINI: lambda: litellm.GeminiModelInfo(),
|
||||
LlmProviders.LITELLM_PROXY: lambda: litellm.LiteLLMProxyChatConfig(),
|
||||
LlmProviders.TOPAZ: lambda: litellm.TopazModelInfo(),
|
||||
LlmProviders.ANTHROPIC: lambda: litellm.AnthropicModelInfo(),
|
||||
LlmProviders.XAI: lambda: litellm.XAIModelInfo(),
|
||||
LlmProviders.OPPER: lambda: litellm.OpperConfig(),
|
||||
LlmProviders.LEMONADE: lambda: litellm.LemonadeChatConfig(),
|
||||
LlmProviders.CLARIFAI: lambda: litellm.ClarifaiConfig(),
|
||||
}
|
||||
)
|
||||
|
||||
@staticmethod
|
||||
def _build_provider_config_map() -> dict[LlmProviders, tuple[Callable, bool]]:
|
||||
"""Build the provider-to-config mapping dictionary.
|
||||
|
|
@ -8888,39 +8907,24 @@ class ProviderConfigManager:
|
|||
model: str | None,
|
||||
provider: LlmProviders,
|
||||
) -> BaseLLMModelInfo | None:
|
||||
if LlmProviders.FIREWORKS_AI == provider:
|
||||
return litellm.FireworksAIConfig()
|
||||
elif LlmProviders.OPENAI == provider:
|
||||
return litellm.OpenAIGPTConfig()
|
||||
elif LlmProviders.GEMINI == provider:
|
||||
return litellm.GeminiModelInfo()
|
||||
elif LlmProviders.VERTEX_AI == provider:
|
||||
factory: Final = ProviderConfigManager._MODEL_INFO_FACTORIES.get(provider)
|
||||
if factory is not None:
|
||||
return factory()
|
||||
if LlmProviders.VERTEX_AI == provider:
|
||||
from litellm.llms.vertex_ai.common_utils import VertexAIModelInfo
|
||||
|
||||
return VertexAIModelInfo()
|
||||
elif LlmProviders.LITELLM_PROXY == provider:
|
||||
return litellm.LiteLLMProxyChatConfig()
|
||||
elif LlmProviders.TOPAZ == provider:
|
||||
return litellm.TopazModelInfo()
|
||||
elif LlmProviders.ANTHROPIC == provider:
|
||||
return litellm.AnthropicModelInfo()
|
||||
elif LlmProviders.XAI == provider:
|
||||
return litellm.XAIModelInfo()
|
||||
elif LlmProviders.OLLAMA == provider or LlmProviders.OLLAMA_CHAT == provider:
|
||||
elif provider in (LlmProviders.OLLAMA, LlmProviders.OLLAMA_CHAT):
|
||||
# Dynamic model listing for Ollama server
|
||||
from litellm.llms.ollama.common_utils import OllamaModelInfo
|
||||
|
||||
return OllamaModelInfo()
|
||||
elif LlmProviders.VLLM == provider or LlmProviders.HOSTED_VLLM == provider:
|
||||
elif provider in (LlmProviders.VLLM, LlmProviders.HOSTED_VLLM):
|
||||
from litellm.llms.vllm.common_utils import (
|
||||
VLLMModelInfo, # experimental approach, to reduce bloat on __init__.py
|
||||
)
|
||||
|
||||
return VLLMModelInfo()
|
||||
elif LlmProviders.LEMONADE == provider:
|
||||
return litellm.LemonadeChatConfig()
|
||||
elif LlmProviders.CLARIFAI == provider:
|
||||
return litellm.ClarifaiConfig()
|
||||
elif LlmProviders.BEDROCK == provider:
|
||||
from litellm.llms.bedrock.common_utils import BedrockModelInfo
|
||||
|
||||
|
|
|
|||
|
|
@ -1,13 +1,36 @@
|
|||
import json
|
||||
import os
|
||||
from unittest.mock import Mock, patch
|
||||
from unittest.mock import patch
|
||||
|
||||
import httpx
|
||||
import pytest
|
||||
|
||||
OPPER_API_BASE = "https://api.opper.ai/v3/compat"
|
||||
|
||||
|
||||
@pytest.fixture(autouse=True)
|
||||
def _without_ambient_opper_credentials(monkeypatch):
|
||||
"""A developer's own OPPER_* environment must not decide where these tests point."""
|
||||
monkeypatch.delenv("OPPER_API_KEY", raising=False)
|
||||
monkeypatch.delenv("OPPER_API_BASE", raising=False)
|
||||
|
||||
|
||||
def _completion_body(usage: dict) -> dict:
|
||||
return {
|
||||
"id": "chatcmpl-123",
|
||||
"object": "chat.completion",
|
||||
"created": 1677652288,
|
||||
"model": "anthropic/claude-haiku-4-5",
|
||||
"choices": [
|
||||
{
|
||||
"index": 0,
|
||||
"message": {"role": "assistant", "content": "Hello!"},
|
||||
"finish_reason": "stop",
|
||||
}
|
||||
],
|
||||
"usage": usage,
|
||||
}
|
||||
|
||||
|
||||
def test_opper_provider_registered():
|
||||
import litellm
|
||||
|
||||
|
|
@ -54,120 +77,48 @@ def test_opper_chat_config_registered():
|
|||
assert isinstance(config, litellm.OpperConfig)
|
||||
|
||||
|
||||
def test_opper_cost_tracking_non_streaming():
|
||||
"""usage.cost from the response body is surfaced as the provider-reported response cost."""
|
||||
@pytest.mark.respx()
|
||||
def test_opper_cost_tracking_non_streaming(respx_mock, monkeypatch):
|
||||
"""usage.cost from the gateway body is surfaced as the provider-reported response cost."""
|
||||
import litellm
|
||||
from litellm.llms.openai.chat.gpt_transformation import OpenAIGPTConfig
|
||||
from litellm.types.utils import Choices, Message, ModelResponse, Usage
|
||||
|
||||
config = litellm.OpperConfig()
|
||||
monkeypatch.setattr(litellm, "disable_aiohttp_transport", True)
|
||||
|
||||
mock_response = Mock(spec=httpx.Response)
|
||||
mock_response.json.return_value = {
|
||||
"id": "chatcmpl-123",
|
||||
"model": "anthropic/claude-haiku-4-5",
|
||||
"choices": [
|
||||
{
|
||||
"message": {"role": "assistant", "content": "Hello!"},
|
||||
"finish_reason": "stop",
|
||||
"index": 0,
|
||||
}
|
||||
],
|
||||
"usage": {
|
||||
"prompt_tokens": 10,
|
||||
"completion_tokens": 20,
|
||||
"total_tokens": 30,
|
||||
"cost": 3.3e-05,
|
||||
},
|
||||
}
|
||||
mock_response.headers = {}
|
||||
|
||||
model_response = ModelResponse(
|
||||
id="chatcmpl-123",
|
||||
choices=[
|
||||
Choices(
|
||||
finish_reason="stop",
|
||||
index=0,
|
||||
message=Message(content="Hello!", role="assistant"),
|
||||
)
|
||||
],
|
||||
created=1234567890,
|
||||
model="anthropic/claude-haiku-4-5",
|
||||
object="chat.completion",
|
||||
usage=Usage(prompt_tokens=10, completion_tokens=20, total_tokens=30),
|
||||
respx_mock.post(f"{OPPER_API_BASE}/chat/completions").respond(
|
||||
status_code=200,
|
||||
json=_completion_body({"prompt_tokens": 10, "completion_tokens": 20, "total_tokens": 30, "cost": 3.3e-05}),
|
||||
)
|
||||
|
||||
with patch.object(OpenAIGPTConfig, "transform_response", return_value=model_response):
|
||||
result = config.transform_response(
|
||||
model="anthropic/claude-haiku-4-5",
|
||||
raw_response=mock_response,
|
||||
model_response=model_response,
|
||||
logging_obj=Mock(),
|
||||
request_data={},
|
||||
messages=[{"role": "user", "content": "Hello"}],
|
||||
optional_params={},
|
||||
litellm_params={},
|
||||
encoding=None,
|
||||
)
|
||||
|
||||
assert result._hidden_params["additional_headers"]["llm_provider-x-litellm-response-cost"] == 3.3e-05
|
||||
|
||||
|
||||
def test_opper_missing_cost_does_not_fail_response():
|
||||
import litellm
|
||||
from litellm.llms.openai.chat.gpt_transformation import OpenAIGPTConfig
|
||||
from litellm.types.utils import Choices, Message, ModelResponse, Usage
|
||||
|
||||
config = litellm.OpperConfig()
|
||||
|
||||
mock_response = Mock(spec=httpx.Response)
|
||||
mock_response.json.return_value = {"usage": {"prompt_tokens": 1, "completion_tokens": 1, "total_tokens": 2}}
|
||||
mock_response.headers = {}
|
||||
|
||||
model_response = ModelResponse(
|
||||
id="chatcmpl-123",
|
||||
choices=[
|
||||
Choices(
|
||||
finish_reason="stop",
|
||||
index=0,
|
||||
message=Message(content="Hello!", role="assistant"),
|
||||
)
|
||||
],
|
||||
created=1234567890,
|
||||
model="anthropic/claude-haiku-4-5",
|
||||
object="chat.completion",
|
||||
usage=Usage(prompt_tokens=1, completion_tokens=1, total_tokens=2),
|
||||
response = litellm.completion(
|
||||
model="opper/anthropic/claude-haiku-4-5",
|
||||
messages=[{"role": "user", "content": "Hello"}],
|
||||
api_key="test-key",
|
||||
)
|
||||
|
||||
with patch.object(OpenAIGPTConfig, "transform_response", return_value=model_response):
|
||||
result = config.transform_response(
|
||||
model="anthropic/claude-haiku-4-5",
|
||||
raw_response=mock_response,
|
||||
model_response=model_response,
|
||||
logging_obj=Mock(),
|
||||
request_data={},
|
||||
messages=[{"role": "user", "content": "Hello"}],
|
||||
optional_params={},
|
||||
litellm_params={},
|
||||
encoding=None,
|
||||
)
|
||||
assert response._hidden_params["additional_headers"]["llm_provider-x-litellm-response-cost"] == 3.3e-05
|
||||
|
||||
assert "llm_provider-x-litellm-response-cost" not in result._hidden_params.get("additional_headers", {})
|
||||
|
||||
mock_response.json.side_effect = ValueError("malformed body")
|
||||
with patch.object(OpenAIGPTConfig, "transform_response", return_value=model_response):
|
||||
result = config.transform_response(
|
||||
model="anthropic/claude-haiku-4-5",
|
||||
raw_response=mock_response,
|
||||
model_response=model_response,
|
||||
logging_obj=Mock(),
|
||||
request_data={},
|
||||
messages=[{"role": "user", "content": "Hello"}],
|
||||
optional_params={},
|
||||
litellm_params={},
|
||||
encoding=None,
|
||||
)
|
||||
assert "llm_provider-x-litellm-response-cost" not in result._hidden_params.get("additional_headers", {})
|
||||
@pytest.mark.respx()
|
||||
def test_opper_missing_cost_does_not_fail_response(respx_mock, monkeypatch):
|
||||
"""A gateway body without usage.cost still returns content, with no cost claimed."""
|
||||
import litellm
|
||||
|
||||
monkeypatch.setattr(litellm, "disable_aiohttp_transport", True)
|
||||
|
||||
respx_mock.post(f"{OPPER_API_BASE}/chat/completions").respond(
|
||||
status_code=200,
|
||||
json=_completion_body({"prompt_tokens": 1, "completion_tokens": 1, "total_tokens": 2}),
|
||||
)
|
||||
|
||||
response = litellm.completion(
|
||||
model="opper/anthropic/claude-haiku-4-5",
|
||||
messages=[{"role": "user", "content": "Hello"}],
|
||||
api_key="test-key",
|
||||
)
|
||||
|
||||
assert response.choices[0].message.content == "Hello!"
|
||||
additional_headers = response._hidden_params.get("additional_headers") or {}
|
||||
assert "llm_provider-x-litellm-response-cost" not in additional_headers
|
||||
|
||||
|
||||
def test_opper_streaming_requests_usage_by_default():
|
||||
|
|
@ -206,11 +157,11 @@ def test_opper_streaming_requests_usage_by_default():
|
|||
|
||||
|
||||
@pytest.mark.respx()
|
||||
def test_opper_streaming_dispatch_carries_cost(respx_mock):
|
||||
def test_opper_streaming_dispatch_carries_cost(respx_mock, monkeypatch):
|
||||
"""usage.cost from the terminal usage chunk survives dispatch into stream assembly."""
|
||||
import litellm
|
||||
|
||||
litellm.disable_aiohttp_transport = True
|
||||
monkeypatch.setattr(litellm, "disable_aiohttp_transport", True)
|
||||
|
||||
mock_chunks = [
|
||||
"data: "
|
||||
|
|
@ -282,11 +233,11 @@ def test_opper_streaming_dispatch_carries_cost(respx_mock):
|
|||
|
||||
|
||||
@pytest.mark.respx()
|
||||
def test_opper_reasoning_effort_passes_through(respx_mock):
|
||||
def test_opper_reasoning_effort_passes_through(respx_mock, monkeypatch):
|
||||
"""reasoning_effort is not rejected client-side and reaches the request body."""
|
||||
import litellm
|
||||
|
||||
litellm.disable_aiohttp_transport = True
|
||||
monkeypatch.setattr(litellm, "disable_aiohttp_transport", True)
|
||||
|
||||
config = litellm.OpperConfig()
|
||||
assert "reasoning_effort" in config.get_supported_openai_params(model="anthropic/claude-haiku-4-5")
|
||||
|
|
@ -409,11 +360,11 @@ def test_opper_supported_params_dispatch():
|
|||
|
||||
|
||||
@pytest.mark.respx()
|
||||
def test_opper_unknown_model_raises_not_found(respx_mock):
|
||||
def test_opper_unknown_model_raises_not_found(respx_mock, monkeypatch):
|
||||
"""A 404 from the gateway maps to a non-retryable NotFoundError."""
|
||||
import litellm
|
||||
|
||||
litellm.disable_aiohttp_transport = True
|
||||
monkeypatch.setattr(litellm, "disable_aiohttp_transport", True)
|
||||
|
||||
respx_mock.post(f"{OPPER_API_BASE}/chat/completions").respond(
|
||||
status_code=404,
|
||||
|
|
@ -434,13 +385,13 @@ def test_opper_unknown_model_raises_not_found(respx_mock):
|
|||
|
||||
|
||||
@pytest.mark.respx()
|
||||
def test_opper_async_completion_and_streaming(respx_mock):
|
||||
def test_opper_async_completion_and_streaming(respx_mock, monkeypatch):
|
||||
"""acompletion works non-streaming and streaming through the async dispatch path."""
|
||||
import asyncio
|
||||
|
||||
import litellm
|
||||
|
||||
litellm.disable_aiohttp_transport = True
|
||||
monkeypatch.setattr(litellm, "disable_aiohttp_transport", True)
|
||||
|
||||
respx_mock.post(f"{OPPER_API_BASE}/chat/completions").respond(
|
||||
status_code=200,
|
||||
|
|
@ -517,3 +468,111 @@ def test_opper_async_completion_and_streaming(respx_mock):
|
|||
stream, chunks = asyncio.run(consume_stream())
|
||||
assert any(c.choices and c.choices[0].delta.content == "Hi" for c in chunks)
|
||||
assert any(getattr(c, "usage", None) is not None and c.usage.cost == 1.1e-05 for c in stream.chunks)
|
||||
|
||||
|
||||
@pytest.mark.respx()
|
||||
def test_opper_get_models_prefixes_the_live_catalog(respx_mock):
|
||||
"""The model list is fetched from the gateway, so it tracks the catalog instead of a static table."""
|
||||
import litellm
|
||||
|
||||
route = respx_mock.get(f"{OPPER_API_BASE}/models").respond(
|
||||
status_code=200,
|
||||
json={
|
||||
"object": "list",
|
||||
"data": [
|
||||
{"id": "anthropic/claude-haiku-4-5", "object": "model"},
|
||||
{"id": "openai/gpt-5.6", "object": "model"},
|
||||
],
|
||||
},
|
||||
)
|
||||
|
||||
models = litellm.OpperConfig().get_models(api_key="test-key")
|
||||
|
||||
assert models == ["opper/anthropic/claude-haiku-4-5", "opper/openai/gpt-5.6"]
|
||||
assert route.calls.last.request.headers["Authorization"] == "Bearer test-key"
|
||||
|
||||
|
||||
@pytest.mark.respx()
|
||||
def test_opper_get_models_keeps_the_gateway_path(respx_mock):
|
||||
"""The catalog lives under the gateway path, not at the host root like plain OpenAI."""
|
||||
import litellm
|
||||
|
||||
route = respx_mock.get("https://gw.example.com/v3/compat/models").respond(
|
||||
status_code=200,
|
||||
json={"object": "list", "data": [{"id": "mistral/mistral-large", "object": "model"}]},
|
||||
)
|
||||
|
||||
models = litellm.OpperConfig().get_models(api_key="test-key", api_base="https://gw.example.com/v3/compat/")
|
||||
|
||||
assert models == ["opper/mistral/mistral-large"]
|
||||
assert route.called
|
||||
|
||||
|
||||
@pytest.mark.respx()
|
||||
def test_opper_get_models_raises_on_error_status(respx_mock):
|
||||
import litellm
|
||||
|
||||
respx_mock.get(f"{OPPER_API_BASE}/models").respond(
|
||||
status_code=401,
|
||||
json={"error": "missing authorization header"},
|
||||
)
|
||||
|
||||
with pytest.raises(ValueError, match="401"):
|
||||
litellm.OpperConfig().get_models(api_key="bad-key")
|
||||
|
||||
|
||||
def test_opper_get_models_requires_an_api_key():
|
||||
import litellm
|
||||
|
||||
with pytest.raises(ValueError, match="OPPER_API_KEY"):
|
||||
litellm.OpperConfig().get_models()
|
||||
|
||||
|
||||
@pytest.mark.respx()
|
||||
def test_opper_valid_models_come_from_the_provider_endpoint(respx_mock):
|
||||
"""get_valid_models reaches the gateway catalog through the registered provider config."""
|
||||
import litellm
|
||||
|
||||
api_base = "https://valid-models.example.com/v3/compat"
|
||||
respx_mock.get(f"{api_base}/models").respond(
|
||||
status_code=200,
|
||||
json={
|
||||
"object": "list",
|
||||
"data": [
|
||||
{"id": "anthropic/claude-sonnet-5", "object": "model"},
|
||||
{"id": "gemini/gemini-3-flash-preview", "object": "model"},
|
||||
],
|
||||
},
|
||||
)
|
||||
|
||||
models = litellm.get_valid_models(
|
||||
check_provider_endpoint=True,
|
||||
custom_llm_provider="opper",
|
||||
api_key="test-key",
|
||||
api_base=api_base,
|
||||
)
|
||||
|
||||
assert models == ["opper/anthropic/claude-sonnet-5", "opper/gemini/gemini-3-flash-preview"]
|
||||
|
||||
|
||||
def test_opper_is_selectable_in_the_add_model_form():
|
||||
from pathlib import Path
|
||||
|
||||
import litellm
|
||||
|
||||
path = Path(litellm.__file__).parent / "proxy" / "public_endpoints" / "provider_create_fields.json"
|
||||
with open(path) as f:
|
||||
entries = [e for e in json.load(f) if e["litellm_provider"] == "opper"]
|
||||
|
||||
assert len(entries) == 1, "opper must appear exactly once in provider_create_fields.json"
|
||||
|
||||
entry = entries[0]
|
||||
assert entry["provider"] == "Opper"
|
||||
assert entry["provider_display_name"] == "Opper"
|
||||
assert entry["default_model_placeholder"].startswith("opper/")
|
||||
|
||||
fields = {f["key"]: f for f in entry["credential_fields"]}
|
||||
assert fields["api_key"]["required"] is True
|
||||
assert fields["api_key"]["field_type"] == "password"
|
||||
assert fields["api_base"]["required"] is False
|
||||
assert fields["api_base"]["placeholder"] == OPPER_API_BASE
|
||||
|
|
|
|||
|
|
@ -6455,3 +6455,41 @@ def test_completion_finishes_response_metadata_before_handing_the_response_to_th
|
|||
assert snapshot["litellm_call_id"]
|
||||
assert snapshot["response_cost"] is not None
|
||||
assert snapshot["api_base"]
|
||||
|
||||
|
||||
@pytest.mark.parametrize(
|
||||
"provider, expected_class_name",
|
||||
[
|
||||
(LlmProviders.FIREWORKS_AI, "FireworksAIConfig"),
|
||||
(LlmProviders.OPENAI, "OpenAIGPTConfig"),
|
||||
(LlmProviders.GEMINI, "GeminiModelInfo"),
|
||||
(LlmProviders.VERTEX_AI, "VertexAIModelInfo"),
|
||||
(LlmProviders.LITELLM_PROXY, "LiteLLMProxyChatConfig"),
|
||||
(LlmProviders.TOPAZ, "TopazModelInfo"),
|
||||
(LlmProviders.ANTHROPIC, "AnthropicModelInfo"),
|
||||
(LlmProviders.XAI, "XAIModelInfo"),
|
||||
(LlmProviders.OPPER, "OpperConfig"),
|
||||
(LlmProviders.OLLAMA, "OllamaModelInfo"),
|
||||
(LlmProviders.OLLAMA_CHAT, "OllamaModelInfo"),
|
||||
(LlmProviders.VLLM, "VLLMModelInfo"),
|
||||
(LlmProviders.HOSTED_VLLM, "VLLMModelInfo"),
|
||||
(LlmProviders.LEMONADE, "LemonadeChatConfig"),
|
||||
(LlmProviders.CLARIFAI, "ClarifaiConfig"),
|
||||
(LlmProviders.BEDROCK, "BedrockModelInfo"),
|
||||
(LlmProviders.AZURE_AI, "AzureFoundryModelInfo"),
|
||||
],
|
||||
)
|
||||
def test_get_provider_model_info_resolves_every_mapped_provider(provider, expected_class_name):
|
||||
"""Every provider with dynamic model info keeps resolving to its own class."""
|
||||
from litellm.utils import ProviderConfigManager
|
||||
|
||||
model_info = ProviderConfigManager.get_provider_model_info(model="some-model", provider=provider)
|
||||
|
||||
assert model_info is not None
|
||||
assert type(model_info).__name__ == expected_class_name
|
||||
|
||||
|
||||
def test_get_provider_model_info_returns_none_for_unmapped_provider():
|
||||
from litellm.utils import ProviderConfigManager
|
||||
|
||||
assert ProviderConfigManager.get_provider_model_info(model="some-model", provider=LlmProviders.DEEPSEEK) is None
|
||||
|
|
|
|||
1
ui/litellm-dashboard/public/assets/logos/opper.svg
Normal file
1
ui/litellm-dashboard/public/assets/logos/opper.svg
Normal file
|
|
@ -0,0 +1 @@
|
|||
<svg viewBox="0 0 315 315" fill="none" xmlns="http://www.w3.org/2000/svg"><g transform="translate(49.5 0)"><path d="M159.78 315C71.53 315 0 244.49 0 157.5C0 -18.9499 159.78 0.650075 159.78 0.650075C159.78 87.2201 88.36 157.4 0.2 157.5C149.8 157.64 159.78 315 159.78 315ZM160.52 217.98C160.52 217.98 156.94 161.65 105.04 157.52C120.6 157.34 160.52 151.54 160.52 96.5601C160.52 151.54 200.44 157.34 216 157.52C164.1 161.63 160.52 217.98 160.52 217.98Z" fill="#1B2E40"/></g></svg>
|
||||
|
After Width: | Height: | Size: 478 B |
|
|
@ -73,6 +73,17 @@ describe("provider_info_helpers", () => {
|
|||
expect(fromEnumKey.logo).toBe(providerLogoMap[Providers.SCX_AI]);
|
||||
});
|
||||
|
||||
it("should map the opper slug and Opper enum key to the Opper display name and logo", () => {
|
||||
const fromSlug = getProviderLogoAndName("opper");
|
||||
expect(fromSlug.displayName).toBe(Providers.Opper);
|
||||
expect(fromSlug.logo).toBe(providerLogoMap[Providers.Opper]);
|
||||
expect(fromSlug.logo).toBeTruthy();
|
||||
|
||||
const fromEnumKey = getProviderLogoAndName("Opper");
|
||||
expect(fromEnumKey.displayName).toBe(Providers.Opper);
|
||||
expect(fromEnumKey.logo).toBe(providerLogoMap[Providers.Opper]);
|
||||
});
|
||||
|
||||
it("should map bedrock_mantle slug to Bedrock Mantle display name and logo", () => {
|
||||
const result = getProviderLogoAndName("bedrock_mantle");
|
||||
expect(result.displayName).toBe(Providers.BedrockMantle);
|
||||
|
|
@ -205,6 +216,10 @@ describe("provider_info_helpers", () => {
|
|||
expect(getPlaceholder(Providers.SCX_AI)).toBe("scx-ai/GLM-5.2");
|
||||
});
|
||||
|
||||
it("should return an opper model placeholder for Opper provider", () => {
|
||||
expect(getPlaceholder(Providers.Opper)).toBe("opper/anthropic/claude-haiku-4-5");
|
||||
});
|
||||
|
||||
it("should return claude-3-opus placeholder for Anthropic provider", () => {
|
||||
expect(getPlaceholder(Providers.Anthropic)).toBe("claude-3-opus");
|
||||
});
|
||||
|
|
|
|||
|
|
@ -43,6 +43,7 @@ import nvidiaTritonLogo from "../../public/assets/logos/nvidia_triton.png";
|
|||
import ollamaLogo from "../../public/assets/logos/ollama.svg";
|
||||
import openaiSmallLogo from "../../public/assets/logos/openai_small.svg";
|
||||
import openrouterLogo from "../../public/assets/logos/openrouter.svg";
|
||||
import opperLogo from "../../public/assets/logos/opper.svg";
|
||||
import oracleLogo from "../../public/assets/logos/oracle.svg";
|
||||
import perplexityAiLogo from "../../public/assets/logos/perplexity-ai.svg";
|
||||
import qwenLogo from "../../public/assets/logos/qwen.png";
|
||||
|
|
@ -145,6 +146,7 @@ export enum Providers {
|
|||
OpenAI_Text = "OpenAI Text Completion",
|
||||
OpenAI_Text_Compatible = "OpenAI-Compatible Completions (legacy /v1/completions)",
|
||||
Openrouter = "Openrouter",
|
||||
Opper = "Opper",
|
||||
Oracle = "Oracle Cloud Infrastructure (OCI)",
|
||||
OVHCLOUD = "Ovhcloud",
|
||||
Perplexity = "Perplexity",
|
||||
|
|
@ -260,6 +262,7 @@ export const provider_map: Record<string, string> = {
|
|||
OpenAI_Text: "text-completion-openai",
|
||||
OpenAI_Text_Compatible: "text-completion-openai",
|
||||
Openrouter: "openrouter",
|
||||
Opper: "opper",
|
||||
Oracle: "oci",
|
||||
OVHCLOUD: "ovhcloud",
|
||||
Perplexity: "perplexity",
|
||||
|
|
@ -362,6 +365,7 @@ export const providerLogoMap: Partial<Record<Providers, string>> = {
|
|||
[Providers.OpenAI_Text_Compatible]: openaiSmallLogo.src,
|
||||
[Providers.OpenAI_Compatible]: openaiSmallLogo.src,
|
||||
[Providers.Openrouter]: openrouterLogo.src,
|
||||
[Providers.Opper]: opperLogo.src,
|
||||
[Providers.Oracle]: oracleLogo.src,
|
||||
[Providers.Perplexity]: perplexityAiLogo.src,
|
||||
[Providers.Qwen_AI_Platform]: qwenLogo.src,
|
||||
|
|
@ -436,6 +440,7 @@ const providerPlaceholderMap: Partial<Record<Providers, string>> = {
|
|||
[Providers.Google_AI_Studio]: "gemini-pro",
|
||||
[Providers.JinaAI]: "jina_ai/",
|
||||
[Providers.NVIDIA_RIVA]: "nvidia_riva/nvidia/parakeet-ctc-1_1b-asr",
|
||||
[Providers.Opper]: "opper/anthropic/claude-haiku-4-5",
|
||||
[Providers.Oracle]: "oci/xai.grok-4",
|
||||
[Providers.RunwayML]: "runwayml/gen4_turbo",
|
||||
[Providers.SageMaker]: "sagemaker/jumpstart-dft-meta-textgeneration-llama-2-7b",
|
||||
|
|
|
|||
Loading…
Add table
Reference in a new issue