[Misc] remove legacy full_duplex alias for auto_response

[NickCao]

Purpose

Keep auto_response as the sole runtime switch for automatic duplex responses.

Test Plan

vLLM Version: 0.30.0

vLLM-Omni Commit: f6f748e

grep for full_duplex

Test Result

no reference remains in either code or doc.

[vllm-omni-review-bot]

This PR appears to belong to: docs/design/module/model_integration.md, docs/design/module/ar_runtime.md, docs/design/module/engine_orchestration.md.

Module owners: @tzhouam @fake0fan @Gaohan123

Routing: @tzhouam via module of the changed files, semantic router, CODEOWNERS; @fake0fan via module of the changed files, semantic router; @Gaohan123 via module of the changed files

@NickCao, please review your own changes and leave a short self-review comment describing what you checked. PRs without author self-review may not be assigned a reviewer.

Please take a look when you have a chance. If you would like an automated review, mention @vllm-omni-review-bot in a comment.

[NickCao]

@vllm-omni-review-bot

[vllm-omni-review-bot]

Omni ReviewBot triage note

Automated triage of commit ed9806f494cd produced:

• Priority: high. Prompt maintainer attention is suggested.

These are automated triage suggestions only — the final decision belongs to the maintainers.

[vllm-omni-review-bot]

Omni ReviewBot routing record

Assigned Strict under experiment vllm-omni-strict-5050-20260829.

[vllm-omni-review-bot]

Omni ReviewBot review

Scan:

Category	Result
Tests / verification	2 finding(s) below
Security	no finding reported
Docs / comments	2 finding(s) below
Behavior / compatibility	no finding reported
Correctness	1 finding(s) below

Validated:

• [resolved] Bot self-review/routing chatter — not a code defect at this head.
• [claim-verified] Purpose: sole runtime switch is auto_response — emitter.py:128, serving_adapter.py:120, plugin.py:163 all drop the full_duplex OR.
• [validated] in-repo producers already on auto_response only: vllm_omni/clients/duplex.py:282 {"auto_response": self.auto_response}; examples realtime_web profiles; docs/serving/realtime_duplex_api.md:641
• [validated] gate_report clean; no CI intersection ask.
• [validated] No new external dependency calls; checklist 10 N/A.
• [validated] public docs never listed the alias: realtime_duplex_api.md:466 extra_body keys are (auto_response, force_listen_count, duplex_initial_user_text)

Keep three items on the primary alias removal: (7) shared auto_responds() now silently ignores full_duplex-only clients while Nemotron still fails loud — demoted to minor and merged with the docs tombstone; (0) no tests pin the new negative contract — merge of the adversary/investigator test asks; (10) Test Result overclaims “no full_duplex in code or doc” — demoted to nit. Drop no_issue/resolved/excluded notes and optional _require_native_full_duplex renames; dup Nemotron/Qwen/site-local repeats onto 0 or 7.

Verdict: COMMENT

Findings

• **[P2] This diff makes auto-response / Nemotron native full-duplex depend only on ext…** — vllm_omni/engine/duplex/session/emitter.pyThis diff makes auto-response / Nemotron native full-duplex depend only onextra.get("auto_response") is True(and Qwen only rejects that key). The formerfull_duplexOR-alias is gone. No test files are in the PR, and existing duplex tests only setauto_responseTrue/False — none assertextra_body={"full_duplex": True}. Add a focused negative unit that full_duplex: Truealone does not makeSessionEmitter.auto_responds()True and does not satisfy Nemotron_require_native_full_duplex` (optional: Qwen no longer raises on that key alone), so reintroducing the alias cannot regress unnoticed.

Evidence: emitter.py:128 return extra.get("auto_response") is True; serving_adapter.py:120 enabled = isinstance(extra_body, dict) and extra_body.get("auto_response") is True; plugin.py:163 if extra_body.get("auto_response"):; unchanged by this diff, present in the PR-time tree: clients/duplex.py:243 auto_response: bool = True; unchanged by this diff: tests/ grep for ['"]full_duplex['"] / extra_body.*full_duplex returns zero matches — duplex harnesses only pass auto_response (e.g. tests/engine/duplex/test_session_runner.py:222 body: dict[str, object] = {"auto_response": auto_response, **(extra_body or {})}).

• **[P2] Qwen validate_client_extra_body no longer rejects full_duplex (only auto_r…** — `` Qwen validate_client_extra_bodyno longer rejectsfull_duplex(onlyauto_response/realtime_tools); with the alias deleted, full_duplex` is an ignored unknown extra — expected residual.

Evidence: vllm_omni/model_executor/models/qwen3_omni/duplex/plugin.py:163-166 if extra_body.get("auto_response"): / raise DuplexRuntimeConfigError( / "Qwen requires server VAD or explicit commits; native auto_response is unsupported" — full_duplex gate removed; no full_duplex matches remain under vllm_omni/model_executor/models/qwen3_omni/ (PR-time tree grep).

• [P2] concern 'docs still teach full_duplex as extra_body key' — not present at head — ``
  concern 'docs still teach full_duplex as extra_body key' — not present at head; docs only brand the Full Duplex feature/pages

Evidence: [resolved] concern 'docs still teach full_duplex as extra_body key' — not present at head; docs only brand the Full Duplex feature/pages

• **[P3] This PR removes only the extra_body wire-key alias (auto_response OR full_…** — docs/serving/full_duplex_api.mdThis PR removes only theextra_body wire-key alias (auto_responseORfull_duplex). Any Test Result that says a grep left “no reference … in either code or doc” overclaims: full_duplexstill appears indocs/serving/full_duplex_api.md(title/prose) and as_require_native_full_duplexinserving_adapter.py. Tighten to “no extra_body.full_duplex alias key remains” (rg ["']full_duplex["']` is empty).

Evidence: unchanged by this diff, present in the PR-time tree: docs/serving/full_duplex_api.md:1 # Full-Duplex WebSocket API; unchanged-by-this-diff name, present in the PR-time tree: vllm_omni/model_executor/models/nemotron_voicechat/duplex/serving_adapter.py:118 def _require_native_full_duplex(config: object) -> None:; post-diff body at serving_adapter.py:120 enabled = isinstance(extra_body, dict) and extra_body.get("auto_response") is True; PR-time tree: rg ["']full_duplex["'] → no matches.

[vllm-omni-review-bot]

[P2] This hunk drops or extra.get("full_duplex") is True from shared `SessionEmitt…

This hunk drops or extra.get("full_duplex") is True from shared SessionEmitter.auto_responds(), so extra_body with only full_duplex: true now yields False with no error on duplex paths that call it. Sibling validators still allow that key through: MiniCPM/PersonaPlex only intersect private-key sets that omit full_duplex, and Aura only type-checks the object — those sessions open. Nemotron alone raises via _require_native_full_duplex when auto_response is True is missing. Alias-only clients therefore get a silent behavior change on non-Nemotron models. Reject leftover full_duplex at session open (or in auto_responds) with an explicit migration error, or document that legacy full_duplex is unrecognized and callers must use auto_response.

Evidence: emitter.py:128 return extra.get("auto_response") is True — full_duplex no longer enables auto-respond. Unchanged by this diff, present in the PR-time tree: minicpmo_4_5/duplex/plugin.py:724-731 private_keys = sorted(PRIVATE_RUNTIME_CONFIG_KEYS.intersection(extra_body)) / raise only on private keys (frozenset at :79-97 omits full_duplex). Unchanged by this diff: personaplex/duplex/serving_adapter.py:151-156 same private-key-only pattern (_PRIVATE_RUNTIME_CONFIG_KEYS at :31-40 omits full_duplex). Unchanged by this diff: aura_omni/duplex/plugin.py:441-445 only rejects non-dict extra_body. Diff-touched: nemotron_voicechat/duplex/serving_adapter.py:120-124 enabled = isinstance(extra_body, dict) and extra_body.get("auto_response") is True then raises unless enabled.

Suggestion: if "full_duplex" in extra:
raise ValueError(
"extra_body.full_duplex is removed; set extra_body.auto_response=true"
)
return extra.get("auto_response") is True

[linyueqian]

Removing the full_duplex alias leaves auto_response as the one switch, and the three sites here are the only remaining readers on main (emitter, Nemotron VoiceChat adapter, Qwen3-Omni duplex plugin), so nothing keeps honouring the old key. Two behaviour notes worth a line in the release notes: a client that still sends extra_body={"full_duplex": true} now silently gets turn-based responses instead of an error, and Qwen3-Omni, which used to reject full_duplex with a DuplexRuntimeConfigError, now accepts the key and ignores it. Static read at f6f748ed against merge-base 7d6e2ade; no PR code executed.

[vllm-omni-review-bot]

Omni ReviewBot: CI is red on this head

@NickCao required checks failed on f6f748edf43b:

• buildkite/vllm-omni

Please fix the failure and push again; this note is updated in place when the head goes green or moves.