GPT-Realtime-2
GPT-Realtime-2 is a speech-to-speech model with configurable reasoning and tools. It remains a distinct release beneath the newer Realtime 2.1 models.
More room for conversation and reasoning
Its 128,000-token window and 32,000-token output ceiling are larger than Realtime 1.5. Text, image and audio inputs share the session, while the output can be text or speech. More reasoning can increase both delay and token use.GPT-Realtime-2 official model documentation
Compare complete voice sessions
Practical guidance. For a multi-step support task, include the tool calls and the final resolution in the comparison. A short initial response can conceal a slow tool cycle. Review whether the assistant uses the latest tool result before explaining the outcome.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Access | API |
| Model identifiers | gpt-realtime-2 |
| Context window | 128,000 tokens Published model context; input and output share this budget GPT-Realtime-2 official model documentation |
| Maximum output | 32,000 tokens Published maximum output; bounded by the remaining context GPT-Realtime-2 official model documentation |
| Input and output | text, audio, image in; text, audio outGPT-Realtime-2 official model documentation |
| API routes | Realtime (v1/realtime)GPT-Realtime-2 official model documentation |
Related models
- GPT-Realtime-2.1
OpenAI · GPT Realtime
GPT-Realtime-2.1 combines conversational audio with reasoning and tool use.
Launch date not confirmedAPI - GPT-Realtime-1.5
OpenAI · GPT Realtime
GPT-Realtime-1.5 supports interactive voice agents with text, audio and image input.
Launch date not confirmedAPI - GPT-Realtime-2.1 Mini
OpenAI · GPT Realtime
GPT-Realtime-2.1 Mini is a distilled voice-reasoning model for lower-cost interactive audio.
Launch date not confirmedAPI - GPT-Realtime-Translate
OpenAI · GPT Realtime
GPT-Realtime-Translate converts incoming speech into translated speech and text while the source audio is still arriving.
Launch date not confirmedAPI