Tracker›Changelog›2026-09-25-request-conversation-mode-und-6-removed
Instrument state: Schema 22 · v2 · Bookmarklet 2.7 · measure · gpt-5-6
{
"send_payload_keys": [
"conversation_mode",
"enable_message_followups",
"force_parallel_switch",
"model_response_contracts",
"paragen_cot_summary_display_override",
"supports_buffering",
"system_hints"
]
}{
"har.send_payload_keys": [
"action",
"client_contextual_info",
"client_prepare_state",
"is_do_not_remember",
"local_function_names",
"messages",
"model",
"supported_encodings",
"timezone",
"timezone_offset_min"
]
}request/conversation-mode request/enable-message-followups request/force-parallel-switch request/model-response-contracts request/paragen-cot-summary-display-override request/supports-buffering request/system-hints
For tools: One of 17 keys in POST /f/conversation (har.send_payload_keys / send_values). Constant „auto" for us. Alongside it, among others, model, paragen_cot_summary_display_override „allow", client_prepare_state „success", supports_buffering.
For visibility: No proven effect. A client parameter of the send request, not a retrieval signal. Relevant to tool builders because changes to the payload structure can break send logic.
| 25 Sept 2026 | observed |
What does force_parallel_switch toggle when it is not „auto"? Parallel tool calls? Never observed in another value.