Responses API: undocumented reasoning+message pairing constraint breaks multi-turn conversations

[achandmsft]

Confirm this is a Node library issue and not an underlying OpenAI API issue

• This is an issue with the Node library

Describe the bug

Reasoning+message items must appear as consecutive pairs in input, but nothing documents this. Common multi-turn patterns silently produce orphaned items → 400 on the next turn.

This broke OpenClaw (64.9k forks) on gpt-5.3-codex. They had to add downgradeOpenAIReasoningBlocks() to strip orphan reasoning items.

The most common failure: filtering response.output to keep only messages (dropping reasoning) → 400: Item 'msg_...' of type 'message' was provided without its required preceding item of type 'reasoning'.

Tested in JS, Python, Go, Java, .NET — identical results across all 5 SDKs. This is an API-level constraint, not SDK-specific. The SDK types don't prevent building input arrays that violate it.

Model	A (all items)	B (msgs only)
gpt-5.3-codex (reasoning=high)	PASS	FAIL
o4-mini	PASS	FAIL*

* Nondeterministic — o4-mini sometimes returns reasoning-only output (no message), so B passes on those runs since there's no message to orphan. Codex with reasoning=high reliably returns both items, making it a deterministic repro.

Workaround: previous_response_id. For manual history, always pass reasoning+message pairs together.

Related:

• Responses API: undocumented reasoning+message pairing constraint breaks multi-turn conversations across all SDKs openai-openapi#536 (spec root cause)
• Responses API: undocumented reasoning+message pairing constraint breaks multi-turn conversations openai-python#3009 (same bug, Python SDK)
• [BUG] Responses API: undocumented reasoning+message pairing constraint breaks multi-turn conversations openai-dotnet#1050 (same bug, .NET SDK)
• Responses API: undocumented reasoning+message pairing constraint breaks multi-turn conversations openai-go#631 (same bug, Go SDK)
• Responses API: undocumented reasoning+message pairing constraint breaks multi-turn conversations openai-java#710 (same bug, Java SDK)
• openclaw/openclaw#49167 (64.9k forks, production breakage)
• Error "Item ‘rs_ABCD’ of type ‘reasoning’ was provided without its required..." when using CodeInterpreter openai-python#2561 (open since Aug 2025)
• GPT-5 + tool calls: Error code: 400 - Item 'rs_...' of type 'reasoning' was provided without its required following item. openai-agents-python#1660 (Agents SDK)

To Reproduce

import OpenAI from "openai";

const client = new OpenAI();
const conversation: any[] = [];

for (const msg of ["Write a Python prime checker.", "Add type hints.", "Add docstrings."]) {
    conversation.push({ role: "user", content: msg });
    const response = await client.responses.create({
        model: "gpt-5.3-codex",
        input: conversation,
        max_output_tokens: 300,
        reasoning: { effort: "high" },
    });
    // Common pattern: keep only messages, discard reasoning
    for (const item of response.output) {
        if (item.type === "message") {
            conversation.push(item);  // orphan message → 400 on next turn
        }
    }
}
// Turn 2 → 400: Item 'msg_...' was provided without its required preceding item

Reproduced on o4-mini and gpt-5.3-codex.

OS

Windows 11, also reproduced on Linux

Node version

Node v24.x

Library version

openai v6.32.0