|
| 1 | +--- |
| 2 | +title: OpenAI |
| 3 | +description: "Learn about using Sentry for OpenAI." |
| 4 | +--- |
| 5 | + |
| 6 | +This integration connects Sentry with the [OpenAI Python SDK](https://github.com/openai/openai-python). |
| 7 | +The integration has been confirmed to work with OpenAI 1.13.3. |
| 8 | + |
| 9 | +## Install |
| 10 | + |
| 11 | +Install `sentry-sdk` from PyPI with the `openai` extra: |
| 12 | + |
| 13 | +```bash |
| 14 | +pip install --upgrade 'sentry-sdk[openai]' |
| 15 | +``` |
| 16 | + |
| 17 | +## Configure |
| 18 | + |
| 19 | +If you have the `openai` package in your dependencies, the OpenAI integration will be enabled automatically when you initialize the Sentry SDK. |
| 20 | + |
| 21 | +An additional dependency, `tiktoken`, is required if you want to calculate token usage for streaming chat responses. |
| 22 | + |
| 23 | +<SignInNote /> |
| 24 | + |
| 25 | +```python |
| 26 | +from openai import OpenAI |
| 27 | + |
| 28 | +import sentry_sdk |
| 29 | + |
| 30 | +sentry_sdk.init( |
| 31 | + dsn="___PUBLIC_DSN___", |
| 32 | + enable_tracing=True, |
| 33 | + traces_sample_rate=1.0, |
| 34 | +) |
| 35 | + |
| 36 | +client = OpenAI() |
| 37 | +``` |
| 38 | + |
| 39 | +## Verify |
| 40 | + |
| 41 | +Verify that the integration works by inducing an error. The error and performance transaction should appear in your Sentry project. |
| 42 | +```python |
| 43 | +from openai import OpenAI |
| 44 | +import sentry_sdk |
| 45 | + |
| 46 | +sentry_sdk.init(...) # same as above |
| 47 | + |
| 48 | +client = OpenAI(api_key="a bad API key") |
| 49 | +with sentry_sdk.start_transaction(op="ai-inference", name="The result of the AI inference"): |
| 50 | + response = ( |
| 51 | + client.chat.completions.create( |
| 52 | + model="some-model", messages=[{"role": "system", "content": "hello"}] |
| 53 | + ) |
| 54 | + .choices[0] |
| 55 | + .message.content |
| 56 | + ) |
| 57 | + print(response) |
| 58 | +``` |
| 59 | + |
| 60 | +After running this script, a transaction will be created in the Performance section of [sentry.io](https://sentry.io). Additionally, an error event (about the bad API key) will be sent to [sentry.io](https://sentry.io) and will be connected to the transaction. |
| 61 | + |
| 62 | +It takes a couple of moments for the data to appear in [sentry.io](https://sentry.io). |
| 63 | + |
| 64 | +## Behavior |
| 65 | + |
| 66 | +- The OpenAI integration will connect Sentry with all supported OpenAI methods automatically. |
| 67 | + |
| 68 | +- All exceptions leading to an OpenAIException are reported. |
| 69 | + |
| 70 | +- The supported modules are currently `chat.completions.create` and `embeddings.create`. |
| 71 | + |
| 72 | +- Sentry is configured not to consider LLM and tokenizer inputs/outputs as PII. If you want to include them, set `send_default_pii=True` in the `sentry_sdk.init()` call. To explicitly exclude prompts despite `send_default_pii=True`, configure the integration with `include_prompts=False` like in the Options section. |
| 73 | + |
| 74 | +## Options |
| 75 | + |
| 76 | +By adding `OpenAIIntegration` to your `sentry_sdk.init()` call explicitly, you can set options for `OpenAIIntegration` to change its behavior: |
| 77 | + |
| 78 | +```python |
| 79 | +import sentry_sdk |
| 80 | +from sentry_sdk.integrations.openai import OpenAIIntegration |
| 81 | + |
| 82 | +sentry_sdk.init( |
| 83 | + dsn="___PUBLIC_DSN___", |
| 84 | + enable_tracing=True, |
| 85 | + send_default_pii=True, |
| 86 | + traces_sample_rate=1.0, |
| 87 | + integrations = [ |
| 88 | + OpenAIIntegration( |
| 89 | + include_prompts=False, # LLM/tokenizer inputs/outputs will be not sent to Sentry, despite send_default_pii=True |
| 90 | + ), |
| 91 | + ], |
| 92 | +) |
| 93 | +``` |
| 94 | + |
| 95 | +## Supported Versions |
| 96 | + |
| 97 | +- OpenAI: 1.0+ |
| 98 | +- tiktoken: 0.6.0+ |
| 99 | +- Python: 3.9+ |
0 commit comments