[gpt-oss][1b] streaming add item id, content id

[qandrew]

Purpose

• implement the FIXMEs, so we now track item_id and content_id, as defined in OpenAI spec (ie https://platform.openai.com/docs/api-reference/responses_streaming/response/content_part/added)
• added a unit test

Test Plan

curl http://localhost:20001/v1/responses   -H "Content-Type: application/json"   -N   -d '{
    "model": "/data/users/axia/checkpoints/gpt-oss-120b",
    "input": [
        {
            "role": "user",
            "content": "Hello."
        }
    ],
    "temperature": 0.7,
    "max_output_tokens": 256,
    "stream": true
}'

Test Result

Note: the item id, content_index changes as expected

event: response.output_item.added
data: {"item":{"id":"msg_81a71aae8e5f4af0a7c5d2f0e6bc6324","summary":[],"type":"reasoning","content":null,"encrypted_content":null,"status":"in_progress"},"output_index":0,"sequence_number":2,"type":"response.output_item.added"}

event: response.content_part.added
data: {"content_index":0,"item_id":"msg_81a71aae8e5f4af0a7c5d2f0e6bc6324","output_index":0,"part":{"annotations":[],"text":"","type":"output_text","logprobs":[]},"sequence_number":3,"type":"response.content_part.added"}

event: response.reasoning_text.delta
data: {"content_index":0,"delta":"The","item_id":"msg_81a71aae8e5f4af0a7c5d2f0e6bc6324","output_index":0,"sequence_number":4,"type":"response.reasoning_text.delta"}

event: response.output_item.done
data: {"item":{"id":"msg_81a71aae8e5f4af0a7c5d2f0e6bc6324","summary":[],"type":"reasoning","content":[{"text":"The user says \"Hello.\" Likely they want a greeting. We respond politely.","type":"reasoning_text"}],"encrypted_content":null,"status":"completed"},"output_index":1,"sequence_number":22,"type":"response.output_item.done"}

event: response.output_item.added
data: {"item":{"id":"msg_c8b805f6ff3b42948a1debd89a2961eb","content":[],"role":"assistant","status":"in_progress","type":"message"},"output_index":1,"sequence_number":23,"type":"response.output_item.added"}

event: response.content_part.added
data: {"content_index":1,"item_id":"msg_c8b805f6ff3b42948a1debd89a2961eb","output_index":1,"part":{"annotations":[],"text":"","type":"output_text","logprobs":[]},"sequence_number":24,"type":"response.content_part.added"}

event: response.output_text.delta
data: {"content_index":1,"delta":"Hello","item_id":"msg_c8b805f6ff3b42948a1debd89a2961eb","logprobs":[],"output_index":1,"sequence_number":25,"type":"response.output_text.delta"}

event: response.output_text.delta
data: {"content_index":1,"delta":"!","item_id":"msg_c8b805f6ff3b42948a1debd89a2961eb","logprobs":[],"output_index":1,"sequence_number":26,"type":"response.output_text.delta"}

event: response.output_text.delta
data: {"content_index":1,"delta":" How","item_id":"msg_c8b805f6ff3b42948a1debd89a2961eb","logprobs":[],"output_index":1,"sequence_number":27,"type":"response.output_text.delta"}

...

OAI example:

ResponseCreatedEvent(response=Response(id='resp_68bf41ccb7f881a3b89e1bbc39dca02008013d49ada9bc0a', created_at=1757364684.0, error=None, incomplete_details=None, instructions=None, metadata={}, model='gpt-5-2025-08-07', object='response', output=[], parallel_tool_calls=True, temperature=1.0, tool_choice='auto', tools=[], top_p=1.0, background=False, conversation=None, max_output_tokens=None, max_tool_calls=None, previous_response_id=None, prompt=None, prompt_cache_key=None, reasoning=Reasoning(effort='medium', generate_summary=None, summary='detailed'), safety_identifier=None, service_tier='auto', status='in_progress', text=ResponseTextConfig(format=ResponseFormatText(type='text'), verbosity='medium'), top_logprobs=0, truncation='disabled', usage=None, user=None, store=True), sequence_number=0, type='response.created')
ResponseInProgressEvent(response=Response(id='resp_68bf41ccb7f881a3b89e1bbc39dca02008013d49ada9bc0a', created_at=1757364684.0, error=None, incomplete_details=None, instructions=None, metadata={}, model='gpt-5-2025-08-07', object='response', output=[], parallel_tool_calls=True, temperature=1.0, tool_choice='auto', tools=[], top_p=1.0, background=False, conversation=None, max_output_tokens=None, max_tool_calls=None, previous_response_id=None, prompt=None, prompt_cache_key=None, reasoning=Reasoning(effort='medium', generate_summary=None, summary='detailed'), safety_identifier=None, service_tier='auto', status='in_progress', text=ResponseTextConfig(format=ResponseFormatText(type='text'), verbosity='medium'), top_logprobs=0, truncation='disabled', usage=None, user=None, store=True), sequence_number=1, type='response.in_progress')
ResponseOutputItemAddedEvent(item=ResponseReasoningItem(id='rs_68bf41cd40c081a3a772340ccea427bf08013d49ada9bc0a', summary=[], type='reasoning', content=None, encrypted_content=None, status=None), output_index=0, sequence_number=2, type='response.output_item.added')
ResponseReasoningSummaryPartAddedEvent(item_id='rs_68bf41cd40c081a3a772340ccea427bf08013d49ada9bc0a', output_index=0, part=Part(text='', type='summary_text'), sequence_number=3, summary_index=0, type='response.reasoning_summary_part.added')
ResponseReasoningSummaryTextDeltaEvent(delta='**Multip', item_id='rs_68bf41cd40c081a3a772340ccea427bf08013d49ada9bc0a', output_index=0, sequence_number=4, summary_index=0, type='response.reasoning_summary_text.delta', obfuscation='gunpSbbS')
ResponseReasoningSummaryTextDeltaEvent(delta='lying', item_id='rs_68bf41cd40c081a3a772340ccea427bf08013d49ada9bc0a', output_index=0, sequence_number=5, summary_index=0, type='response.reasoning_summary_text.delta', obfuscation='Y5Doun05WUq')
ResponseReasoningSummaryTextDeltaEvent(delta=' with', item_id='rs_68bf41cd40c081a3a772340ccea427bf08013d49ada9bc0a', output_index=0, sequence_number=6, summary_index=0, type='response.reasoning_summary_text.delta', obfuscation='fIXeF6P3dpv')
ResponseReasoningSummaryTextDeltaEvent(delta=' precision', item_id='rs_68bf41cd40c081a3a772340ccea427bf08013d49ada9bc0a', output_index=0, sequence_number=7, summary_index=0, type='response.reasoning_summary_text.delta', obfuscation='EXuwuv')
...
ResponseReasoningSummaryTextDoneEvent(item_id='rs_68bf41cd40c081a3a772340ccea427bf08013d49ada9bc0a', output_index=0, sequence_number=105, summary_index=0, text='**Multiplying with precision**\n\nI need to multiply 3.2342 by 233.1123123. The user likely meant "multiply" when they wrote "multiple," so I’ll compute the product accurately. \n\nI\'ll start by using standard multiplication methods and breaking down the calculations into parts. First, I\'ll calculate b * 3 and then b * 0.2342. After summing up everything, I arrive at approximately 753.9318. Now I should double-check my work to ensure the accuracy of this result.', type='response.reasoning_summary_text.done')
ResponseReasoningSummaryPartDoneEvent(item_id='rs_68bf41cd40c081a3a772340ccea427bf08013d49ada9bc0a', output_index=0, part=Part(text='**Multiplying with precision**\n\nI need to multiply 3.2342 by 233.1123123. The user likely meant "multiply" when they wrote "multiple," so I’ll compute the product accurately. \n\nI\'ll start by using standard multiplication methods and breaking down the calculations into parts. First, I\'ll calculate b * 3 and then b * 0.2342. After summing up everything, I arrive at approximately 753.9318. Now I should double-check my work to ensure the accuracy of this result.', type='summary_text'), sequence_number=106, summary_index=0, type='response.reasoning_summary_part.done')
ResponseReasoningSummaryPartAddedEvent(item_id='rs_68bf41cd40c081a3a772340ccea427bf08013d49ada9bc0a', output_index=0, part=Part(text='', type='summary_text'), sequence_number=107, summary_index=1, type='response.reasoning_summary_part.added')
ResponseReasoningSummaryTextDeltaEvent(delta='**Ver', item_id='rs_68bf41cd40c081a3a772340ccea427bf08013d49ada9bc0a', output_index=0, sequence_number=108, summary_index=1, type='response.reasoning_summary_text.delta', obfuscation='fAEej8BrrXI')
ResponseReasoningSummaryTextDeltaEvent(delta='ifying', item_id='rs_68bf41cd40c081a3a772340ccea427bf08013d49ada9bc0a', output_index=0, sequence_number=109, summary_index=1, type='response.reasoning_summary_text.delta', obfuscation='y3VTtzA8kO')
ResponseReasoningSummaryTextDeltaEvent(delta=' multiplication', item_id='rs_68bf41cd40c081a3a772340ccea427bf08013d49ada9bc0a', output_index=0, sequence_number=110, summary_index=1, type='response.reasoning_summary_text.delta', obfuscation='O')
ResponseReasoningSummaryTextDeltaEvent(delta=' accuracy', item_id='rs_68bf41cd40c081a3a772340ccea427bf08013d49ada9bc0a', output_index=0, sequence_number=111, summary_index=1, type='response.reasoning_summary_text.delta', obfuscation='8EOQcwh')
ResponseReasoningSummaryTextDeltaEvent(delta='**\n\nI', item_id='rs_68bf41cd40c081a3a772340ccea427bf08013d49ada9bc0a', output_index=0, sequence_number=112, summary_index=1, type='response.reasoning_summary_text.delta', obfuscation='PNZJtO5JRJd')
ResponseReasoningSummaryTextDeltaEvent(delta='’m', item_id='rs_68bf41cd40c081a3a772340ccea427bf08013d49ada9bc0a', output_index=0, sequence_number=113, summary_index=1, type='response.reasoning_summary_text.delta', obfuscation='bD9H4GV0ImvosW')
ResponseReasoningSummaryTextDeltaEvent(delta=' considering', item_id='rs_68bf41cd40c081a3a772340ccea427bf08013d49ada9bc0a', output_index=0, sequence_number=114, summary_index=1, type='response.reasoning_summary_text.delta', obfuscation='q7tS')
ResponseReasoningSummaryTextDeltaEvent(delta=' using', item_id='rs_68bf41cd40c081a3a772340ccea427bf08013d49ada9bc0a', output_index=0, sequence_number=115, summary_index=1, type='response.reasoning_summary_text.delta', obfuscation='ozdIa1egB2')
ResponseReasoningSummaryTextDeltaEvent(delta=' Python', item_id='rs_68bf41cd40c081a3a772340ccea427bf08013d49ada9bc0a', output_index=0, sequence_number=116, summary_index=1, type='response.reasoning_summary_text.delta', obfuscation='ZqZS6kqzp')
...
ResponseReasoningSummaryPartDoneEvent(item_id='rs_68bf41cd40c081a3a772340ccea427bf08013d49ada9bc0a', output_index=0, part=Part(text='**Determining significant digits**\n\nI can confirm that the multiplication of \\( 3.2342 \\) (4 decimal places) and \\( 233.1123123 \\) (7 decimal places) results in a number with at most 11 decimal places. My computed result, \\( 753.93184044066 \\), indeed has 11 digits after the decimal. \n\nI should correct the user’s phrasing of "multiple" to "multiply." It’s good to be concise, so I’ll present the product clearly, maybe with a brief explanation or a one-liner to keep it straightforward. Let\'s go ahead and show that computed result!', type='summary_text'), sequence_number=371, summary_index=2, type='response.reasoning_summary_part.done')
ResponseOutputItemDoneEvent(item=ResponseReasoningItem(id='rs_68bf41cd40c081a3a772340ccea427bf08013d49ada9bc0a', summary=[Summary(text='**Multiplying with precision**\n\nI need to multiply 3.2342 by 233.1123123. The user likely meant "multiply" when they wrote "multiple," so I’ll compute the product accurately. \n\nI\'ll start by using standard multiplication methods and breaking down the calculations into parts. First, I\'ll calculate b * 3 and then b * 0.2342. After summing up everything, I arrive at approximately 753.9318. Now I should double-check my work to ensure the accuracy of this result.', type='summary_text'), Summary(text='**Verifying multiplication accuracy**\n\nI’m considering using Python for high-precision calculations of \\( a \\times b \\). First, I’ll confirm the previous result: \\( 3.2342 \\times 233.1123123 \\). I recalled that \\( 233.1123123 \\times 3 \\) equals approximately \\( 699.3369369 \\), which checks out. \n\nNow, I’ll look closely at calculating \\( 0.2342 \\times b \\). A direct computation gives around \\( 54.59490354066 \\), and it matches my previous sum. The total is accurately \\( 753.93184044066 \\), which I’ll present as the final rounded result.', type='summary_text'), Summary(text='**Determining significant digits**\n\nI can confirm that the multiplication of \\( 3.2342 \\) (4 decimal places) and \\( 233.1123123 \\) (7 decimal places) results in a number with at most 11 decimal places. My computed result, \\( 753.93184044066 \\), indeed has 11 digits after the decimal. \n\nI should correct the user’s phrasing of "multiple" to "multiply." It’s good to be concise, so I’ll present the product clearly, maybe with a brief explanation or a one-liner to keep it straightforward. Let\'s go ahead and show that computed result!', type='summary_text')], type='reasoning', content=None, encrypted_content=None, status=None), output_index=0, sequence_number=372, type='response.output_item.done')
ResponseOutputItemAddedEvent(item=ResponseOutputMessage(id='msg_68bf41e9d4f481a3a52a8532f944df3808013d49ada9bc0a', content=[], role='assistant', status='in_progress', type='message'), output_index=1, sequence_number=373, type='response.output_item.added')
ResponseContentPartAddedEvent(content_index=0, item_id='msg_68bf41e9d4f481a3a52a8532f944df3808013d49ada9bc0a', output_index=1, part=ResponseOutputText(annotations=[], text='', type='output_text', logprobs=[]), sequence_number=374, type='response.content_part.added')
ResponseTextDeltaEvent(content_index=0, delta='3', item_id='msg_68bf41e9d4f481a3a52a8532f944df3808013d49ada9bc0a', logprobs=[], output_index=1, sequence_number=375, type='response.output_text.delta', obfuscation='5UpTHBZr8ovDHaN')
ResponseTextDeltaEvent(content_index=0, delta='.', item_id='msg_68bf41e9d4f481a3a52a8532f944df3808013d49ada9bc0a', logprobs=[], output_index=1, sequence_number=376, type='response.output_text.delta', obfuscation='M130VTwll5a8aEc')
ResponseTextDeltaEvent(content_index=0, delta='234', item_id='msg_68bf41e9d4f481a3a52a8532f944df3808013d49ada9bc0a', logprobs=[], output_index=1, sequence_number=377, type='response.output_text.delta', obfuscation='An2UEHoufSxRq')
ResponseTextDeltaEvent(content_index=0, delta='2', item_id='msg_68bf41e9d4f481a3a52a8532f944df3808013d49ada9bc0a', logprobs=[], output_index=1, sequence_number=378, type='response.output_text.delta', obfuscation='Lt6gv4l7d2FzIxv')

---

Essential Elements of an Effective PR Description Checklist

• The purpose of the PR, such as "Fix some issue (link existing issues this PR will resolve)".
• The test plan, such as providing test command.
• The test results, such as pasting the results comparison before and after, or e2e results
• (Optional) The necessary documentation update, such as updating supported_models.md and examples for a new model.
• (Optional) Release notes update. If your change is user facing, please update the release notes draft in the Google Doc.

[yeqcharlotte]

this test case doesn't capture multiple subsequent streaming items?

[qandrew]

hmm i'm not sure if i understand your question? I haven't enabled the tool calling for streaming yet, so currently we're only testing reasoningOutput -> finalOutput items.

[chaunceyjiang]

Could you provide a sample response from the OpenAI online service?

[qandrew]

Could you provide a sample response from the OpenAI online service?

yep, added in the description

[chaunceyjiang]

Thanks~

It looks consistent with OpenAI’s format.