fix(openai): record service_tier attribute (#3458)

ataha322 · web-flow · commit 589e1047edf9 · 2025-11-23T20:37:49.000+02:00
diff --git a/packages/opentelemetry-instrumentation-openai/opentelemetry/instrumentation/openai/shared/__init__.py b/packages/opentelemetry-instrumentation-openai/opentelemetry/instrumentation/openai/shared/__init__.py
@@ -12,6 +12,7 @@
 )
 from opentelemetry.semconv._incubating.attributes import (
     gen_ai_attributes as GenAIAttributes,
+    openai_attributes as OpenAIAttributes,
 )
 from opentelemetry.semconv_ai import SpanAttributes
 from opentelemetry.trace.propagation import set_span_in_context
@@ -141,6 +142,9 @@ def _set_request_attributes(span, kwargs, instance=None):
     _set_span_attribute(
         span, SpanAttributes.LLM_IS_STREAMING, kwargs.get("stream") or False
     )
+    _set_span_attribute(
+        span, OpenAIAttributes.OPENAI_REQUEST_SERVICE_TIER, kwargs.get("service_tier")
+    )
     if response_format := kwargs.get("response_format"):
         # backward-compatible check for
         # openai.types.shared_params.response_format_json_schema.ResponseFormatJSONSchema
@@ -210,6 +214,11 @@ def _set_response_attributes(span, response):
         SpanAttributes.LLM_OPENAI_RESPONSE_SYSTEM_FINGERPRINT,
         response.get("system_fingerprint"),
     )
+    _set_span_attribute(
+        span,
+        OpenAIAttributes.OPENAI_RESPONSE_SERVICE_TIER,
+        response.get("service_tier"),
+    )
     _log_prompt_filter(span, response)
     usage = response.get("usage")
     if not usage:
diff --git a/packages/opentelemetry-instrumentation-openai/opentelemetry/instrumentation/openai/v1/responses_wrappers.py b/packages/opentelemetry-instrumentation-openai/opentelemetry/instrumentation/openai/v1/responses_wrappers.py
@@ -42,6 +42,7 @@
 from opentelemetry.instrumentation.utils import _SUPPRESS_INSTRUMENTATION_KEY
 from opentelemetry.semconv._incubating.attributes import (
     gen_ai_attributes as GenAIAttributes,
+    openai_attributes as OpenAIAttributes,
 )
 from opentelemetry.semconv_ai import SpanAttributes
 from opentelemetry.semconv.attributes.error_attributes import ERROR_TYPE
@@ -132,6 +133,10 @@ class TracedData(pydantic.BaseModel):
     request_reasoning_effort: Optional[str] = pydantic.Field(default=None)
     response_reasoning_effort: Optional[str] = pydantic.Field(default=None)
 
+    # OpenAI service tier
+    request_service_tier: Optional[str] = pydantic.Field(default=None)
+    response_service_tier: Optional[str] = pydantic.Field(default=None)
+
 
 responses: dict[str, TracedData] = {}
 
@@ -189,6 +194,8 @@ def set_data_attributes(traced_response: TracedData, span: Span):
     _set_span_attribute(span, GenAIAttributes.GEN_AI_REQUEST_MODEL, traced_response.request_model)
     _set_span_attribute(span, GenAIAttributes.GEN_AI_RESPONSE_ID, traced_response.response_id)
     _set_span_attribute(span, GenAIAttributes.GEN_AI_RESPONSE_MODEL, traced_response.response_model)
+    _set_span_attribute(span, OpenAIAttributes.OPENAI_REQUEST_SERVICE_TIER, traced_response.request_service_tier)
+    _set_span_attribute(span, OpenAIAttributes.OPENAI_RESPONSE_SERVICE_TIER, traced_response.response_service_tier)
     if usage := traced_response.usage:
         _set_span_attribute(span, GenAIAttributes.GEN_AI_USAGE_INPUT_TOKENS, usage.input_tokens)
         _set_span_attribute(span, GenAIAttributes.GEN_AI_USAGE_OUTPUT_TOKENS, usage.output_tokens)
@@ -483,6 +490,8 @@ def responses_get_or_create_wrapper(tracer: Tracer, wrapped, instance, args, kwa
                     )
                 ),
                 response_reasoning_effort=kwargs.get("reasoning", {}).get("effort"),
+                request_service_tier=kwargs.get("service_tier"),
+                response_service_tier=existing_data.get("response_service_tier"),
             )
         except Exception:
             traced_data = None
@@ -546,6 +555,8 @@ def responses_get_or_create_wrapper(tracer: Tracer, wrapped, instance, args, kwa
                 )
             ),
             response_reasoning_effort=kwargs.get("reasoning", {}).get("effort"),
+            request_service_tier=existing_data.get("request_service_tier", kwargs.get("service_tier")),
+            response_service_tier=existing_data.get("response_service_tier", parsed_response.service_tier),
         )
         responses[parsed_response.id] = traced_data
     except Exception:
@@ -621,6 +632,8 @@ async def async_responses_get_or_create_wrapper(
                     )
                 ),
                 response_reasoning_effort=kwargs.get("reasoning", {}).get("effort"),
+                request_service_tier=kwargs.get("service_tier"),
+                response_service_tier=existing_data.get("response_service_tier"),
             )
         except Exception:
             traced_data = None
@@ -685,6 +698,8 @@ async def async_responses_get_or_create_wrapper(
                 )
             ),
             response_reasoning_effort=kwargs.get("reasoning", {}).get("effort"),
+            request_service_tier=existing_data.get("request_service_tier", kwargs.get("service_tier")),
+            response_service_tier=existing_data.get("response_service_tier", parsed_response.service_tier),
         )
         responses[parsed_response.id] = traced_data
     except Exception:
@@ -791,6 +806,8 @@ def __init__(
             ),
             request_reasoning_effort=self._request_kwargs.get("reasoning", {}).get("effort"),
             response_reasoning_effort=None,
+            request_service_tier=self._request_kwargs.get("service_tier"),
+            response_service_tier=None,
         )
 
         self._complete_response_data = None
diff --git a/packages/opentelemetry-instrumentation-openai/tests/traces/cassettes/test_chat/test_chat_with_service_tier.yaml b/packages/opentelemetry-instrumentation-openai/tests/traces/cassettes/test_chat/test_chat_with_service_tier.yaml
@@ -0,0 +1,113 @@
+interactions:
+- request:
+    body: '{"messages": [{"role": "user", "content": "Say hello"}], "model": "gpt-5",
+      "service_tier": "priority"}'
+    headers:
+      accept:
+      - application/json
+      accept-encoding:
+      - gzip, deflate
+      connection:
+      - keep-alive
+      content-length:
+      - '102'
+      content-type:
+      - application/json
+      host:
+      - api.openai.com
+      traceparent:
+      - 00-9abdb187b56105d7f694cb1bcd7d6ff7-ff31170765953060-01
+      user-agent:
+      - OpenAI/Python 1.99.7
+      x-stainless-arch:
+      - arm64
+      x-stainless-async:
+      - 'false'
+      x-stainless-lang:
+      - python
+      x-stainless-os:
+      - MacOS
+      x-stainless-package-version:
+      - 1.99.7
+      x-stainless-read-timeout:
+      - '600'
+      x-stainless-retry-count:
+      - '0'
+      x-stainless-runtime:
+      - CPython
+      x-stainless-runtime-version:
+      - 3.9.6
+    method: POST
+    uri: https://api.openai.com/v1/chat/completions
+  response:
+    body:
+      string: !!binary |
+        H4sIAAAAAAAAA3RSQW7bMBC86xVbnq1Cduok9iVIU6AuCgQucmsQCCy5tthSXJZcNVUC/70g7Vgy
+        0lx02NkZzQz3uQAQRoslCNVIVq235c1m/vh5vVp//B2fbtSnVf9tzYu72+u7p+9fb8UkMejHT1T8
+        wnqvqPUW2ZDbwyqgZEyq04vzs8W0OpvOMtCSRptoW8/lvJxVs3lZXZbVxYHXkFEYxRLuCwCA5/xN
+        Dp3Gv2IJ1eRl0mKMcotieVwCEIFsmggZo4ksHYvJACpyjC6bXqG19A5W9AhKOvgCDVoPPXXApGV/
+        NaYF3HRRJtOus3YESOeIZQqdDT8ckN3R4sY4E5s6oIzk0m8jkxcZ3RUADzlyd5JC+ECt55rpF2bZ
+        y72aGBoeYbMDyMTSDvPFoaVTsVojS2PjqDGhpGpQD8yhXtlpQyOgGEV7beZ/2vvYxm0HlfMPb+oP
+        gFLoGXXtA2qjThMPawHTAb61duw4OxYRwx+jsGaDIb2DD4aC4X5/dLGPjG29MW6LwQeTTyS9dbEr
+        /gEAAP//AwB9nhCJHwMAAA==
+    headers:
+      CF-RAY:
+      - 9a319038ca178687-ARN
+      Connection:
+      - keep-alive
+      Content-Encoding:
+      - gzip
+      Content-Type:
+      - application/json
+      Date:
+      - Sun, 23 Nov 2025 15:05:13 GMT
+      Server:
+      - cloudflare
+      Set-Cookie:
+      - __cf_bm=BzRM2cFojUtXKFo1i0gAAj1_aXPTRwMVXEDci5ux0eY-1763910313-1.0.1.1-Cq0yyUFvMCbE6GRA747glZtqTPRyXEkAE44E2L5S_TwRwNKZ.r26uFYeXvwQbhTSqDZ2KeyvsS693q_z9Dd9aWBj.j0eUJcBE5nNJ.pIhOg;
+        path=/; expires=Sun, 23-Nov-25 15:35:13 GMT; domain=.api.openai.com; HttpOnly;
+        Secure; SameSite=None
+      - _cfuvid=BoDvu6D4M8n.b5OoTqVk4rwbKfyeViFSKx_hJR4S4W0-1763910313444-0.0.1.1-604800000;
+        path=/; domain=.api.openai.com; HttpOnly; Secure; SameSite=None
+      Strict-Transport-Security:
+      - max-age=31536000; includeSubDomains; preload
+      Transfer-Encoding:
+      - chunked
+      X-Content-Type-Options:
+      - nosniff
+      access-control-expose-headers:
+      - X-Request-ID
+      alt-svc:
+      - h3=":443"; ma=86400
+      cf-cache-status:
+      - DYNAMIC
+      openai-organization:
+      - agentpaid
+      openai-processing-ms:
+      - '968'
+      openai-project:
+      - proj_g54CtVh9wETqz1duiA1cw9Kw
+      openai-version:
+      - '2020-10-01'
+      x-envoy-upstream-service-time:
+      - '1163'
+      x-openai-proxy-wasm:
+      - v0.1
+      x-ratelimit-limit-requests:
+      - '500'
+      x-ratelimit-limit-tokens:
+      - '500000'
+      x-ratelimit-remaining-requests:
+      - '499'
+      x-ratelimit-remaining-tokens:
+      - '499995'
+      x-ratelimit-reset-requests:
+      - 120ms
+      x-ratelimit-reset-tokens:
+      - 0s
+      x-request-id:
+      - req_b623980c393049c6b6d99da2cc70576a
+    status:
+      code: 200
+      message: OK
+version: 1
diff --git a/packages/opentelemetry-instrumentation-openai/tests/traces/cassettes/test_responses/test_responses_with_service_tier.yaml b/packages/opentelemetry-instrumentation-openai/tests/traces/cassettes/test_responses/test_responses_with_service_tier.yaml
@@ -0,0 +1,110 @@
+interactions:
+- request:
+    body: '{"input": "Say hello", "model": "gpt-5", "service_tier": "priority"}'
+    headers:
+      accept:
+      - application/json
+      accept-encoding:
+      - gzip, deflate
+      connection:
+      - keep-alive
+      content-length:
+      - '68'
+      content-type:
+      - application/json
+      host:
+      - api.openai.com
+      user-agent:
+      - OpenAI/Python 1.99.7
+      x-stainless-arch:
+      - arm64
+      x-stainless-async:
+      - 'false'
+      x-stainless-lang:
+      - python
+      x-stainless-os:
+      - MacOS
+      x-stainless-package-version:
+      - 1.99.7
+      x-stainless-read-timeout:
+      - '600'
+      x-stainless-retry-count:
+      - '0'
+      x-stainless-runtime:
+      - CPython
+      x-stainless-runtime-version:
+      - 3.9.6
+    method: POST
+    uri: https://api.openai.com/v1/responses
+  response:
+    body:
+      string: !!binary |
+        H4sIAAAAAAAAAwAAAP//fFTLctswDLznK1ie444efki+9Jp+QyajgSjIYUMJHBJyo+n43zuibD0a
+        uzcJCywXWJB/noSQupJHIR16W0TbAyRJtIV9lMT17hBF+zxJk6RMsnqnsjjf51msMpWmZRKXCtOd
+        fB4oqPyFim801Hoc48ohMFYFDFh82Kd5HKVJEjDPwJ0fahQ11iBjNRaVoD5Ojrp20FWD8TiGtTG6
+        Pcmj+PMkhBDSQo9uqK/wjIYsOvkkxCUko3M0YG1nTAjo9nZKUSGDNn6NenadYk3tKt7AZ0Ed244L
+        pg/8CjKRKRSYNV1DFZpB2cnyZrdJomS3ibJNdLhOKzDKo3gNjYztzEb4xzakW0i2wYa6yjDKs+qA
+        e8gO47gDCfcWRyPAUzsMbIJ81zTg+uHgtxC7PN8T0PjTfxRkOYRFgBTzapdXaR1vY5UlXxU06D2c
+        FtIeOR5ARS1jO09lKWxFe/MDP3mqDgnQtsRw8/D1bQUaOllH5R0kEB2FfEFj6Jt4od9CQSt+inc0
+        VvTUCaYK+h9yKrpcvyYe6cgEbeC99gwtj8lDYkiSFhwYg2a9MOy6cbetw7Omzhe361MEK6aFso4a
+        y4UC9Y7FB/YPMYfDEDW1y4x5Faa7g3VNjkeXKt0110kuVmSonq6Thxq5L3Q1kNcaV1fLoztrhQWP
+        cWmdJqe5l9drTg6XvTI2Fh1wF8Lx9+gaDS5c5dXkGpj/F+6HvOXyyjO6kvxw3tzNJHwc9ztpNfrT
+        MckJmJdBMtlisSLRFLRLja5rFVynKyvtoTS3R6sLqz41oNvVm5E9fw0v3qGpy2BiNddFq07/fYni
+        3T3gHu+0AY+omRjMDCbpNMHOr91ukKEChoH+8nT5CwAA//8DAD1YGBRCBgAA
+    headers:
+      CF-RAY:
+      - 9a3190799a8282d6-ARN
+      Connection:
+      - keep-alive
+      Content-Encoding:
+      - gzip
+      Content-Type:
+      - application/json
+      Date:
+      - Sun, 23 Nov 2025 15:05:24 GMT
+      Server:
+      - cloudflare
+      Set-Cookie:
+      - __cf_bm=5epdftKChRBshn5d9QnvOUIH6jdbbC2Z49fpF_kViyo-1763910324-1.0.1.1-cHXuCjdspu_Vvx.3Ax1PpY9WHMivIEjCSU1BsYYaHZ_cEDxkd62QXyB39dzmzPGQ9zjTm8MVBplGzsr60dhv0gBrBaj9EQkh58nrbutTgVo;
+        path=/; expires=Sun, 23-Nov-25 15:35:24 GMT; domain=.api.openai.com; HttpOnly;
+        Secure; SameSite=None
+      - _cfuvid=gp4oUQF19kcdJq_p9DCnU9cTap7yKDGhmqIT54evW5M-1763910324011-0.0.1.1-604800000;
+        path=/; domain=.api.openai.com; HttpOnly; Secure; SameSite=None
+      Strict-Transport-Security:
+      - max-age=31536000; includeSubDomains; preload
+      Transfer-Encoding:
+      - chunked
+      X-Content-Type-Options:
+      - nosniff
+      alt-svc:
+      - h3=":443"; ma=86400
+      cf-cache-status:
+      - DYNAMIC
+      openai-organization:
+      - agentpaid
+      openai-processing-ms:
+      - '1395'
+      openai-project:
+      - proj_g54CtVh9wETqz1duiA1cw9Kw
+      openai-version:
+      - '2020-10-01'
+      x-envoy-upstream-service-time:
+      - '1398'
+      x-ratelimit-limit-requests:
+      - '500'
+      x-ratelimit-limit-tokens:
+      - '500000'
+      x-ratelimit-remaining-requests:
+      - '499'
+      x-ratelimit-remaining-tokens:
+      - '500000'
+      x-ratelimit-reset-requests:
+      - 120ms
+      x-ratelimit-reset-tokens:
+      - 0s
+      x-request-id:
+      - req_c7a7a91635ad4051892623b7a91f919f
+    status:
+      code: 200
+      message: OK
+version: 1
diff --git a/packages/opentelemetry-instrumentation-openai/tests/traces/test_chat.py b/packages/opentelemetry-instrumentation-openai/tests/traces/test_chat.py
@@ -1497,6 +1497,27 @@ def test_chat_reasoning(instrument_legacy, span_exporter,
     assert span.attributes["gen_ai.usage.reasoning_tokens"] > 0
 
 
+@pytest.mark.vcr
+def test_chat_with_service_tier(instrument_legacy, span_exporter, openai_client):
+    openai_client.chat.completions.create(
+        model="gpt-5",
+        messages=[
+            {
+                "role": "user",
+                "content": "Say hello"
+            }
+        ],
+        service_tier="priority",
+    )
+
+    spans = span_exporter.get_finished_spans()
+    assert len(spans) >= 1
+    span = spans[-1]
+
+    assert span.attributes["openai.request.service_tier"] == "priority"
+    assert span.attributes["openai.response.service_tier"] == "priority"
+
+
 def test_chat_exception(instrument_legacy, span_exporter, openai_client):
     openai_client.api_key = "invalid"
     with pytest.raises(Exception):
diff --git a/packages/opentelemetry-instrumentation-openai/tests/traces/test_responses.py b/packages/opentelemetry-instrumentation-openai/tests/traces/test_responses.py
@@ -29,6 +29,23 @@ def test_responses(
     # assert span.attributes["gen_ai.prompt.0.role"] == "user"
 
 
+@pytest.mark.vcr
+def test_responses_with_service_tier(
+    instrument_legacy, span_exporter: InMemorySpanExporter, openai_client: OpenAI
+):
+    _ = openai_client.responses.create(
+        model="gpt-5",
+        input="Say hello",
+        service_tier="priority",
+    )
+    spans = span_exporter.get_finished_spans()
+    assert len(spans) == 1
+    span = spans[0]
+    assert span.name == "openai.response"
+    assert span.attributes["openai.request.service_tier"] == "priority"
+    assert span.attributes["openai.response.service_tier"] == "priority"
+
+
 @pytest.mark.vcr
 def test_responses_with_input_history(
     instrument_legacy, span_exporter: InMemorySpanExporter, openai_client: OpenAI
