GPT Realtime 2

JSON →
openai multimodal
textaudioimage

The next-generation realtime model with multimodal support and advanced reasoning capabilities.

context window 32K tokens
max output 4K tokens
input price $4 / 1M tokens
output price $16 / 1M tokens
streamingtool-usefunction-callingvisionreasoning
releasedAug 2025
knowledge cutoffJun 2025