Azure OpenAI Service
An Azure service that provides access to OpenAI’s GPT-3 models with enterprise capabilities.
Responses API charges Code Interpreter, even if not used
![](data:image/svg+xml, %3Csvg xmlns='http://www.w3.org/2000/svg' height='64' class='font-weight-bold' style='font: 600 30.11764705882353px "SegoeUI", Arial' width='64'%3E%3Ccircle fill='hsl(208, 45%, 30%)' cx='32' cy='32' r='32' /%3E%3Ctext x='50%25' y='55%25' dominant-baseline='middle' text-anchor='middle' fill='%23FFF' %3EMV%3C/text%3E%3C/svg%3E)
Markus von Staden
5
Reputation points
Hi everyone,
after migrating from the Assistants API to the Responses API, we are encountering unexpected billing behavior:
For every single request, a Code Interpreter call is being charged, even though:
- type is set to "auto"
- no files are provided
- no tool usage is intended
- and importantly: the model does not return any output_item of type code_interpreter.
In other words, the Code Interpreter is never used, yet it is always billed.
We didn't encounter this behavior with the Assistants API before.
If relevant: We manage our own message history. We still have store=true, because the code interpreter won't work otherwise.
Is this a known issue or reproducible on your side? This 10x our costs last month.
Should Code Interpreter usage be billable even when no output_item is returned?
Thanks in advance!
Azure OpenAI Service
{count} votes
-
![](data:image/svg+xml, %3Csvg xmlns='http://www.w3.org/2000/svg' height='64' class='font-weight-bold' style='font: 600 30.11764705882353px "SegoeUI", Arial' width='64'%3E%3Ccircle fill='hsl(204.8, 15%, 29%)' cx='32' cy='32' r='32' /%3E%3Ctext x='50%25' y='55%25' dominant-baseline='middle' text-anchor='middle' fill='%23FFF' %3EMM%3C/text%3E%3C/svg%3E)
Manas Mohanty 13,340 Reputation points Moderator2025-11-17T17:31:15.71+00:00 Sorry for the inconvenience.
If you are setting tool type as code interpreter with mode
"container": {"type": "auto"}the Responses API eagerly creates a container for the turn, even if the model never issues a
code_interpreter_callin its output items. That container creation is billable.Community reports indicate the API may create a new (billable) container per turn until the model finally decides to use the tool and returns a
code_interpreter_callitem exposing thecontainer_id. Only then can the same container be reused; before that, each turn can still be charged.This behavior deviates from Assistants API, where CI sessions were only billed when actually used.
Suggestions
- Don’t include CI unless you need it on that turn
- Only add
tools: [{ "type": "code_interpreter", "container": {"type": "auto"} }]on requests that you expect will actually require Python execution. - For pure chat/QA turns, omit the CI tool completely to prevent eager container creation (and the charge). [community.openai.com]
- Only add
- Use tool-choice to discourage CI
- When supported for your model/SDK, set
tool_choice: {"type":"none"}(or avoid passing any tools) on turns that shouldn’t touch CI. - Note: Some GPT‑5 model/SDK combinations restrict tool choices to
'auto'for certain tools, so this lever is model/SDK‑dependent. Track the SDK issues for current constraints. [github.com]
- When supported for your model/SDK, set
- Gate CI behind intent detection
- Do a first pass with no tools, ask the model “Please go with code interpreter if you need apython tool?” (or do your own heuristic).
- If the answer is yes, issue a second request with CI enabled; if no, continue without CI.
- This “two‑step” keeps CI containers from being created on non-compute turns.
- Prefer explicit reuse once CI is actually used
- After you see a
code_interpreter_callitem, capture thecontainer_idfrom that item and reuse it in subsequent turns (instead of relying onauto). This reduces repeated container setups. (If your SDK doesn’t expose explicit container reuse primitives, track the maintainers’ guidance/workarounds in the issues.) [github.com]container = client.containers.create(name="test-container") response = client.responses.create( model="gpt-4.1", tools=[{ "type": "code_interpreter", "container": container.id }], tool_choice="required", input="use the python tool to calculate what is 4 * 3.82. and then find its square root and then find the square root of that result" )
- After you see a
- Audit your billing and logs
- Correlate turn-level request logs with Code Interpreter line items to quantify where containers are created without a call. This will help you verify the impact of mitigation #1 and #3 over time. (The MS Q&A thread you referenced is consistent with this audit outcome.
Reference used
- https://community.openai.com/t/responses-api-excessive-charges-due-to-repetitive-container-creation-when-using-code-interpreter/1364328
- https://platform.openai.com/docs/guides/tools-code-interpreter
Hope it helps.
Thank you.
- Don’t include CI unless you need it on that turn
-
Markus von Staden 5 Reputation points
2025-11-18T08:27:24.9666667+00:00 Do I understand this correctly: This isn’t a bug but the intended behavior? That’s a massive downgrade from the Assistants API and really, really bad!
-
Manas Mohanty 13,340 Reputation points Moderator
2025-11-18T10:30:08.17+00:00 Sorry for the experience. will check with team internally and update you.
Thank you.
-
Markus von Staden 5 Reputation points
2025-11-18T20:24:48.2166667+00:00 Update for everyone interested:
This really was a bug and at least OpenAI has deployed a fix. I don't know if Azure will deploy a fix too and if we get the falsely billed containers refunded.
Source: https://community.openai.com/t/horrible-code-interpreter-bug/1366600/9 -
Manas Mohanty 13,340 Reputation points Moderator
2025-11-24T11:36:54.91+00:00 Could you help me with contact information requested in private message to create engineering ticket.
Last update from Product group was "Billing and metering of Azure OpenAI is different than OpenAI".
Thank you.
Sign in to comment
Sign in to answer