Realtime speech-to-speech model with configurable reasoning, tool use, and robust voice-agent behavior

Details

FieldValue
idgpt-realtime-2.1
familygpt
modalitytext+audio+image->text+audio
context_length128000
max_input_tokens96000
max_output_tokens32000
knowledge_cutoff2024-09-30
release_date2026-07-06
is_open_weightsno
supports_tool_callyes
supports_reasoningyes
supports_structured_outputno
supports_attachmentyes
input_modalitiestext, audio, image
output_modalitiestext, audio

Providers