scriptoriumadmin/litellm
1
0
Fork 0
mirror of https://github.com/BerriAI/litellm.git synced 2026-09-12 23:01:41 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 88

[Feat] Day-0 Support for OpenAI Re-usable prompts Responses API (#11782)

Browse source 
* add prompt to responses params

* add OpenAI PromptObject

* add prompt param to responses api

* test_get_optional_params_responses_api

* test_openai_responses_litellm_router_with_prompt

* docs Reusable Prompts
This commit is contained in:
Ishaan Jaff 2025-06-16 21:28:50 -07:00 • committed by GitHub
parent 99c2a7fb70
commit 80501b8268
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
6 changed files with 166 additions and 21 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

44
docs/my-website/docs/providers/openai/responses_api.md
View file

@ -207,6 +207,50 @@ print(delete_response)
|----------|---------------------|
| `openai` | [All Responses API parameters are supported](https://github.com/BerriAI/litellm/blob/7c3df984da8e4dff9201e4c5353fdc7a2b441831/litellm/llms/openai/responses/transformation.py#L23) |
### Reusable Prompts
Use the `prompt` parameter to reference a stored prompt template and optionally supply variables.
```python showLineNumbers title="Stored Prompt"
import litellm
response = litellm.responses(
model="openai/o1-pro",
prompt={
"id": "pmpt_abc123",
"version": "2",
"variables": {
"customer_name": "Jane Doe",
"product": "40oz juice box",
},
},
)
print(response)
```
The same parameter is supported when calling the LiteLLM proxy with the OpenAI SDK:
```python showLineNumbers title="Stored Prompt via Proxy"
from openai import OpenAI
client = OpenAI(base_url="http://localhost:4000", api_key="your-api-key")
response = client.responses.create(
model="openai/o1-pro",
prompt={
"id": "pmpt_abc123",
"version": "2",
"variables": {
"customer_name": "Jane Doe",
"product": "40oz juice box",
},
},
)
print(response)
```
## Computer Use
<Tabs>

1
litellm/llms/openai/responses/transformation.py
View file

@ -38,6 +38,7 @@ class OpenAIResponsesAPIConfig(BaseResponsesAPIConfig):
"store",
"background",
"stream",
"prompt",
"temperature",
"text",
"tool_choice",

44
litellm/responses/main.py
View file

@ -15,6 +15,7 @@ from litellm.responses.litellm_completion_transformation.handler import (
)
from litellm.responses.utils import ResponsesAPIRequestUtils
from litellm.types.llms.openai import (
PromptObject,
Reasoning,
ResponseIncludable,
ResponseInputParam,
@ -96,6 +97,7 @@ async def aresponses(
include: Optional[List[ResponseIncludable]] = None,
instructions: Optional[str] = None,
max_output_tokens: Optional[int] = None,
prompt: Optional[PromptObject] = None,
metadata: Optional[Dict[str, Any]] = None,
parallel_tool_calls: Optional[bool] = None,
previous_response_id: Optional[str] = None,
@ -141,6 +143,7 @@ async def aresponses(
include=include,
instructions=instructions,
max_output_tokens=max_output_tokens,
prompt=prompt,
metadata=metadata,
parallel_tool_calls=parallel_tool_calls,
previous_response_id=previous_response_id,
@ -197,6 +200,7 @@ def responses(
include: Optional[List[ResponseIncludable]] = None,
instructions: Optional[str] = None,
max_output_tokens: Optional[int] = None,
prompt: Optional[PromptObject] = None,
metadata: Optional[Dict[str, Any]] = None,
parallel_tool_calls: Optional[bool] = None,
previous_response_id: Optional[str] = None,
@ -255,11 +259,11 @@ def responses(
)
# get provider config
responses_api_provider_config: Optional[
BaseResponsesAPIConfig
] = ProviderConfigManager.get_provider_responses_api_config(
model=model,
provider=litellm.LlmProviders(custom_llm_provider),
responses_api_provider_config: Optional[BaseResponsesAPIConfig] = (
ProviderConfigManager.get_provider_responses_api_config(
model=model,
provider=litellm.LlmProviders(custom_llm_provider),
)
)
local_vars.update(kwargs)
@ -449,11 +453,11 @@ def delete_responses(
raise ValueError("custom_llm_provider is required but passed as None")
# get provider config
responses_api_provider_config: Optional[
BaseResponsesAPIConfig
] = ProviderConfigManager.get_provider_responses_api_config(
model=None,
provider=litellm.LlmProviders(custom_llm_provider),
responses_api_provider_config: Optional[BaseResponsesAPIConfig] = (
ProviderConfigManager.get_provider_responses_api_config(
model=None,
provider=litellm.LlmProviders(custom_llm_provider),
)
)
if responses_api_provider_config is None:
@ -628,11 +632,11 @@ def get_responses(
raise ValueError("custom_llm_provider is required but passed as None")
# get provider config
responses_api_provider_config: Optional[
BaseResponsesAPIConfig
] = ProviderConfigManager.get_provider_responses_api_config(
model=None,
provider=litellm.LlmProviders(custom_llm_provider),
responses_api_provider_config: Optional[BaseResponsesAPIConfig] = (
ProviderConfigManager.get_provider_responses_api_config(
model=None,
provider=litellm.LlmProviders(custom_llm_provider),
)
)
if responses_api_provider_config is None:
@ -784,11 +788,11 @@ def list_input_items(
if custom_llm_provider is None:
raise ValueError("custom_llm_provider is required but passed as None")
responses_api_provider_config: Optional[
BaseResponsesAPIConfig
] = ProviderConfigManager.get_provider_responses_api_config(
model=None,
provider=litellm.LlmProviders(custom_llm_provider),
responses_api_provider_config: Optional[BaseResponsesAPIConfig] = (
ProviderConfigManager.get_provider_responses_api_config(
model=None,
provider=litellm.LlmProviders(custom_llm_provider),
)
)
if responses_api_provider_config is None:

14
litellm/types/llms/openai.py
View file

@ -930,6 +930,19 @@ class ComputerToolParam(TypedDict, total=False):
ALL_RESPONSES_API_TOOL_PARAMS = Union[ToolParam, ComputerToolParam]
class PromptObject(TypedDict, total=False):
"""Reference to a stored prompt template."""
id: Required[str]
"""The unique identifier of the prompt template to use."""
variables: Optional[Dict]
"""Variables to substitute into the prompt template."""
version: Optional[str]
"""Optional version of the prompt template."""
class ResponsesAPIOptionalRequestParams(TypedDict, total=False):
"""TypedDict for Optional parameters supported by the responses API."""
@ -950,6 +963,7 @@ class ResponsesAPIOptionalRequestParams(TypedDict, total=False):
top_p: Optional[float]
truncation: Optional[Literal["auto", "disabled"]]
user: Optional[str]
prompt: Optional[PromptObject]
class ResponsesAPIRequestParams(ResponsesAPIOptionalRequestParams, total=False):

74
tests/llm_responses_api_testing/test_openai_responses_api.py
View file

@ -689,6 +689,80 @@ async def test_openai_responses_litellm_router_with_metadata():
mock_post.assert_called_once()
@pytest.mark.asyncio
async def test_openai_responses_litellm_router_with_prompt():
"""Test that prompt object is passed through the Router for responses API"""
prompt_obj = {
"id": "pmpt_abc123",
"version": "2",
"variables": {"random_variable": "ishaan_from_litellm"},
}
mock_response = {
"id": "resp_123",
"object": "response",
"created_at": 1741476542,
"status": "completed",
"model": "gpt-4o",
"output": [],
"parallel_tool_calls": True,
"usage": {"input_tokens": 0, "output_tokens": 0, "total_tokens": 0},
"text": {"format": {"type": "text"}},
"error": None,
"incomplete_details": None,
"instructions": None,
"metadata": {},
"temperature": 1.0,
"tool_choice": "auto",
"tools": [],
"top_p": 1.0,
"max_output_tokens": None,
"previous_response_id": None,
"reasoning": {"effort": None, "summary": None},
"truncation": "disabled",
"user": None,
}
class MockResponse:
def __init__(self, json_data, status_code):
self._json_data = json_data
self.status_code = status_code
self.text = str(json_data)
def json(self):
return self._json_data
with patch(
"litellm.llms.custom_httpx.http_handler.AsyncHTTPHandler.post",
new_callable=AsyncMock,
) as mock_post:
mock_post.return_value = MockResponse(mock_response, 200)
litellm._turn_on_debug()
router = litellm.Router(
model_list=[
{
"model_name": "gpt4o-special-alias",
"litellm_params": {
"model": "gpt-4o",
"api_key": "fake-key",
},
}
]
)
await router.aresponses(
model="gpt4o-special-alias",
input="Hello",
prompt=prompt_obj,
)
request_body = mock_post.call_args.kwargs["json"]
assert request_body["prompt"] == prompt_obj
mock_post.assert_called_once()
def test_bad_request_bad_param_error():
"""Raise a BadRequestError when an invalid parameter value is provided"""
try:

10
tests/test_litellm/responses/test_responses_utils.py
View file

@ -25,7 +25,11 @@ class TestResponsesAPIRequestUtils:
model = "gpt-4o"
config = OpenAIResponsesAPIConfig()
optional_params = ResponsesAPIOptionalRequestParams(
{"temperature": 0.7, "max_output_tokens": 100}
{
"temperature": 0.7,
"max_output_tokens": 100,
"prompt": {"id": "pmpt_123"},
}
)
# Execute
@ -41,6 +45,8 @@ class TestResponsesAPIRequestUtils:
assert result["temperature"] == 0.7
assert "max_output_tokens" in result
assert result["max_output_tokens"] == 100
assert "prompt" in result
assert result["prompt"] == {"id": "pmpt_123"}
def test_get_optional_params_responses_api_unsupported_param(self):
"""Test that unsupported parameters raise an error"""
@ -68,6 +74,7 @@ class TestResponsesAPIRequestUtils:
params = {
"temperature": 0.7,
"max_output_tokens": 100,
"prompt": {"id": "pmpt_456"},
"invalid_param": "value",
"model": "gpt-4o", # This is not in ResponsesAPIOptionalRequestParams
}
@ -84,6 +91,7 @@ class TestResponsesAPIRequestUtils:
assert "model" not in result
assert result["temperature"] == 0.7
assert result["max_output_tokens"] == 100
assert result["prompt"] == {"id": "pmpt_456"}
def test_decode_previous_response_id_to_original_previous_response_id(self):
"""Test decoding a LiteLLM encoded previous_response_id to the original previous_response_id"""