Skip to content

Update openai spans to be OTel compatible #4563

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 2 commits into
base: antonpirker/openai-overhaul
Choose a base branch
from
Open

Update openai spans to be OTel compatible #4563

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
3096d74
Updated attribute names to be OTel compatible
antonpirker Jul 9, 2025
5dc8107
Updated tests
antonpirker Jul 9, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
9 changes: 7 additions & 2 deletions sentry_sdk/consts.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -372,6 +372,12 @@ class SPANDATA:
Example: "chat"
"""

GEN_AI_RESPONSE_MODEL = "gen_ai.response.model"
"""
Exact model identifier used to generate the response
Example: gpt-4o-mini-2024-07-18
"""

GEN_AI_RESPONSE_TEXT = "gen_ai.response.text"
"""
The model's response text messages.
Expand Down Expand Up @@ -649,6 +655,7 @@ class OP:
FUNCTION_AWS = "function.aws"
FUNCTION_GCP = "function.gcp"
GEN_AI_CHAT = "gen_ai.chat"
GEN_AI_EMBEDDINGS = "gen_ai.embeddings"
GEN_AI_EXECUTE_TOOL = "gen_ai.execute_tool"
GEN_AI_HANDOFF = "gen_ai.handoff"
GEN_AI_INVOKE_AGENT = "gen_ai.invoke_agent"
Expand All @@ -674,8 +681,6 @@ class OP:
MIDDLEWARE_STARLITE = "middleware.starlite"
MIDDLEWARE_STARLITE_RECEIVE = "middleware.starlite.receive"
MIDDLEWARE_STARLITE_SEND = "middleware.starlite.send"
OPENAI_CHAT_COMPLETIONS_CREATE = "ai.chat_completions.create.openai"
OPENAI_EMBEDDINGS_CREATE = "ai.embeddings.create.openai"
HUGGINGFACE_HUB_CHAT_COMPLETIONS_CREATE = (
"ai.chat_completions.create.huggingface_hub"
)
Expand Down
30 changes: 18 additions & 12 deletions sentry_sdk/integrations/openai.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -136,8 +136,8 @@ def _new_chat_completion_common(f, *args, **kwargs):
streaming = kwargs.get("stream")

span = sentry_sdk.start_span(
op=consts.OP.OPENAI_CHAT_COMPLETIONS_CREATE,
name="Chat Completion",
op=consts.OP.GEN_AI_CHAT,
name=f"{consts.OP.GEN_AI_CHAT} {model}",
origin=OpenAIIntegration.origin,
)
span.__enter__()
Expand All @@ -146,16 +146,16 @@ def _new_chat_completion_common(f, *args, **kwargs):

with capture_internal_exceptions():
if should_send_default_pii() and integration.include_prompts:
set_data_normalized(span, SPANDATA.AI_INPUT_MESSAGES, messages)
set_data_normalized(span, SPANDATA.GEN_AI_REQUEST_MESSAGES, messages)

set_data_normalized(span, SPANDATA.AI_MODEL_ID, model)
set_data_normalized(span, SPANDATA.GEN_AI_REQUEST_MODEL, model)
set_data_normalized(span, SPANDATA.AI_STREAMING, streaming)

if hasattr(res, "choices"):
if should_send_default_pii() and integration.include_prompts:
set_data_normalized(
span,
SPANDATA.AI_RESPONSES,
SPANDATA.GEN_AI_RESPONSE_TEXT,
list(map(lambda x: x.message, res.choices)),
)
_calculate_chat_completion_usage(
Expand Down Expand Up @@ -189,7 +189,7 @@ def new_iterator():
)
if should_send_default_pii() and integration.include_prompts:
set_data_normalized(
span, SPANDATA.AI_RESPONSES, all_responses
span, SPANDATA.GEN_AI_RESPONSE_TEXT, all_responses
)
_calculate_chat_completion_usage(
messages,
Expand Down Expand Up @@ -222,7 +222,7 @@ async def new_iterator_async():
)
if should_send_default_pii() and integration.include_prompts:
set_data_normalized(
span, SPANDATA.AI_RESPONSES, all_responses
span, SPANDATA.GEN_AI_RESPONSE_TEXT, all_responses
)
_calculate_chat_completion_usage(
messages,
Expand Down Expand Up @@ -320,24 +320,30 @@ def _new_embeddings_create_common(f, *args, **kwargs):
if integration is None:
return f(*args, **kwargs)

model = kwargs.get("model")

with sentry_sdk.start_span(
op=consts.OP.OPENAI_EMBEDDINGS_CREATE,
description="OpenAI Embedding Creation",
op=consts.OP.GEN_AI_EMBEDDINGS,
name=f"{consts.OP.GEN_AI_EMBEDDINGS} {model}",
origin=OpenAIIntegration.origin,
) as span:
if "input" in kwargs and (
should_send_default_pii() and integration.include_prompts
):
if isinstance(kwargs["input"], str):
set_data_normalized(span, SPANDATA.AI_INPUT_MESSAGES, [kwargs["input"]])
set_data_normalized(
span, SPANDATA.GEN_AI_REQUEST_MESSAGES, [kwargs["input"]]
)
elif (
isinstance(kwargs["input"], list)
and len(kwargs["input"]) > 0
and isinstance(kwargs["input"][0], str)
):
set_data_normalized(span, SPANDATA.AI_INPUT_MESSAGES, kwargs["input"])
set_data_normalized(
span, SPANDATA.GEN_AI_REQUEST_MESSAGES, kwargs["input"]
)
if "model" in kwargs:
set_data_normalized(span, SPANDATA.AI_MODEL_ID, kwargs["model"])
set_data_normalized(span, SPANDATA.GEN_AI_REQUEST_MODEL, kwargs["model"])

response = yield f, args, kwargs

Expand Down
58 changes: 32 additions & 26 deletions tests/integrations/openai/test_openai.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -81,14 +81,17 @@ def test_nonstreaming_chat_completion(
tx = events[0]
assert tx["type"] == "transaction"
span = tx["spans"][0]
assert span["op"] == "ai.chat_completions.create.openai"
assert span["op"] == "gen_ai.chat"

if send_default_pii and include_prompts:
assert "hello" in span["data"][SPANDATA.AI_INPUT_MESSAGES]["content"]
assert "the model response" in span["data"][SPANDATA.AI_RESPONSES]["content"]
assert "hello" in span["data"][SPANDATA.GEN_AI_REQUEST_MESSAGES]["content"]
assert (
"the model response"
in span["data"][SPANDATA.GEN_AI_RESPONSE_TEXT]["content"]
)
else:
assert SPANDATA.AI_INPUT_MESSAGES not in span["data"]
assert SPANDATA.AI_RESPONSES not in span["data"]
assert SPANDATA.GEN_AI_REQUEST_MESSAGES not in span["data"]
assert SPANDATA.GEN_AI_RESPONSE_TEXT not in span["data"]

assert span["measurements"]["ai_completion_tokens_used"]["value"] == 10
assert span["measurements"]["ai_prompt_tokens_used"]["value"] == 20
Expand Down Expand Up @@ -123,14 +126,17 @@ async def test_nonstreaming_chat_completion_async(
tx = events[0]
assert tx["type"] == "transaction"
span = tx["spans"][0]
assert span["op"] == "ai.chat_completions.create.openai"
assert span["op"] == "gen_ai.chat"

if send_default_pii and include_prompts:
assert "hello" in span["data"][SPANDATA.AI_INPUT_MESSAGES]["content"]
assert "the model response" in span["data"][SPANDATA.AI_RESPONSES]["content"]
assert "hello" in span["data"][SPANDATA.GEN_AI_REQUEST_MESSAGES]["content"]
assert (
"the model response"
in span["data"][SPANDATA.GEN_AI_RESPONSE_TEXT]["content"]
)
else:
assert SPANDATA.AI_INPUT_MESSAGES not in span["data"]
assert SPANDATA.AI_RESPONSES not in span["data"]
assert SPANDATA.GEN_AI_REQUEST_MESSAGES not in span["data"]
assert SPANDATA.GEN_AI_RESPONSE_TEXT not in span["data"]

assert span["measurements"]["ai_completion_tokens_used"]["value"] == 10
assert span["measurements"]["ai_prompt_tokens_used"]["value"] == 20
Expand Down Expand Up @@ -216,14 +222,14 @@ def test_streaming_chat_completion(
tx = events[0]
assert tx["type"] == "transaction"
span = tx["spans"][0]
assert span["op"] == "ai.chat_completions.create.openai"
assert span["op"] == "gen_ai.chat"

if send_default_pii and include_prompts:
assert "hello" in span["data"][SPANDATA.AI_INPUT_MESSAGES]["content"]
assert "hello world" in span["data"][SPANDATA.AI_RESPONSES]
assert "hello" in span["data"][SPANDATA.GEN_AI_REQUEST_MESSAGES]["content"]
assert "hello world" in span["data"][SPANDATA.GEN_AI_RESPONSE_TEXT]
else:
assert SPANDATA.AI_INPUT_MESSAGES not in span["data"]
assert SPANDATA.AI_RESPONSES not in span["data"]
assert SPANDATA.GEN_AI_REQUEST_MESSAGES not in span["data"]
assert SPANDATA.GEN_AI_RESPONSE_TEXT not in span["data"]

try:
import tiktoken # type: ignore # noqa # pylint: disable=unused-import
Expand Down Expand Up @@ -312,14 +318,14 @@ async def test_streaming_chat_completion_async(
tx = events[0]
assert tx["type"] == "transaction"
span = tx["spans"][0]
assert span["op"] == "ai.chat_completions.create.openai"
assert span["op"] == "gen_ai.chat"

if send_default_pii and include_prompts:
assert "hello" in span["data"][SPANDATA.AI_INPUT_MESSAGES]["content"]
assert "hello world" in span["data"][SPANDATA.AI_RESPONSES]
assert "hello" in span["data"][SPANDATA.GEN_AI_REQUEST_MESSAGES]["content"]
assert "hello world" in span["data"][SPANDATA.GEN_AI_RESPONSE_TEXT]
else:
assert SPANDATA.AI_INPUT_MESSAGES not in span["data"]
assert SPANDATA.AI_RESPONSES not in span["data"]
assert SPANDATA.GEN_AI_REQUEST_MESSAGES not in span["data"]
assert SPANDATA.GEN_AI_RESPONSE_TEXT not in span["data"]

try:
import tiktoken # type: ignore # noqa # pylint: disable=unused-import
Expand Down Expand Up @@ -403,11 +409,11 @@ def test_embeddings_create(
tx = events[0]
assert tx["type"] == "transaction"
span = tx["spans"][0]
assert span["op"] == "ai.embeddings.create.openai"
assert span["op"] == "gen_ai.embeddings"
if send_default_pii and include_prompts:
assert "hello" in span["data"][SPANDATA.AI_INPUT_MESSAGES]
assert "hello" in span["data"][SPANDATA.GEN_AI_REQUEST_MESSAGES]
else:
assert SPANDATA.AI_INPUT_MESSAGES not in span["data"]
assert SPANDATA.GEN_AI_REQUEST_MESSAGES not in span["data"]

assert span["measurements"]["ai_prompt_tokens_used"]["value"] == 20
assert span["measurements"]["ai_total_tokens_used"]["value"] == 30
Expand Down Expand Up @@ -451,11 +457,11 @@ async def test_embeddings_create_async(
tx = events[0]
assert tx["type"] == "transaction"
span = tx["spans"][0]
assert span["op"] == "ai.embeddings.create.openai"
assert span["op"] == "gen_ai.embeddings"
if send_default_pii and include_prompts:
assert "hello" in span["data"][SPANDATA.AI_INPUT_MESSAGES]
assert "hello" in span["data"][SPANDATA.GEN_AI_REQUEST_MESSAGES]
else:
assert SPANDATA.AI_INPUT_MESSAGES not in span["data"]
assert SPANDATA.GEN_AI_REQUEST_MESSAGES not in span["data"]

assert span["measurements"]["ai_prompt_tokens_used"]["value"] == 20
assert span["measurements"]["ai_total_tokens_used"]["value"] == 30
Expand Down
Loading