# GPT-Realtime-1.5

GPT-Realtime-1.5 powers production voice agents. One model listens and speaks, follows instructions reliably, and calls your tools mid-conversation, without a separate transcription and synthesis chain.

- **Model ID:** `openai/gpt-realtime-1.5`
- **Type:** realtime
- **Providers:** openai
- **Context window:** Not applicable
- **Maximum output tokens:** 0
- **Pricing:** $4/1M input tokens, $16/1M output tokens
- **Canonical page:** https://vercel.com/ai-gateway/models/gpt-realtime-1.5

## Supported parameters

Detailed capability metadata has not been reported for this model.

## Example

```ts
import { gateway } from '@ai-sdk/gateway';

export async function POST() {
  const { token, url } = await gateway.experimental_realtime.getToken({
    model: 'openai/gpt-realtime-1.5',
  });

  return Response.json({ token, url, tools: [] });
}
```

## About

GPT-Realtime-1.5 launched on February 23, 2026 as OpenAI's flagship realtime audio model for voice agents and customer support. Compared with earlier realtime preview models, GPT-Realtime-1.5 delivers stronger instruction following, more reliable tool calling, and better voice quality.

Speech-to-speech is the defining design choice. Rather than chaining transcription, a text model, and speech synthesis, GPT-Realtime-1.5 processes audio natively and responds in audio. That cuts latency and produces responses that sound more natural. Tool calling works during live conversation, so a voice agent can check a calendar or query a database while the caller is still talking.

GPT-Realtime-1.5 also enacts specified roles reliably, following instructions to hold a particular personality or tone. That consistency makes GPT-Realtime-1.5 practical for branded support experiences and scripted workflows at scale.

## What to consider

Realtime support on AI Gateway is in beta through AI SDK 7. The integration is WebSocket-based: your server mints a short-lived token, and the browser connects with the AI SDK's realtime hook, which handles microphone capture and playback. Your AI Gateway API key never reaches the client.

## When to use

### Best For

- Customer support voice agents that follow scripts and escalate correctly
- Phone-style assistants that call tools and APIs mid-conversation
- Voice interfaces where a chained transcription and synthesis pipeline adds too much latency
- Role-consistent agents that need to hold a defined personality or tone

### Consider Alternatives When

- Conversations require deeper reasoning; `gpt-realtime-2` adds configurable reasoning effort
- You're transcribing recorded audio rather than conversing; use `gpt-4o-transcribe` or `gpt-4o-mini-transcribe`
- Voice features are high volume and cost sensitive; `gpt-realtime-mini` covers simpler interactions

## Best for

- **Customer support agents:** Script-following voice agents that answer, resolve, and escalate correctly
- **Tool-calling voice workflows:** Assistants that check calendars, query databases, or call APIs mid-conversation
- **Low-latency voice interfaces:** Native speech-to-speech avoids the delay of chained transcription and synthesis
- **Role-consistent personas:** Agents that hold a defined personality or tone across long sessions

## Consider alternatives

- **Deeper reasoning required:** `gpt-realtime-2` adds configurable reasoning effort for complex voice workflows
- **Recorded audio transcription:** `gpt-4o-transcribe` and `gpt-4o-mini-transcribe` are built for speech-to-text jobs
- **Cost-sensitive voice features:** `gpt-realtime-mini` handles high-volume, simpler interactions at a lower rate

## Frequently asked questions

### How do I access GPT-Realtime-1.5 through AI Gateway?

Realtime support is in beta through AI SDK 7. Mint a short-lived token on your server with the AI SDK's gateway provider, then connect from the browser with the AI SDK's realtime hook over WebSocket. The hook manages microphone capture and audio playback.

### What does speech-to-speech mean for GPT-Realtime-1.5?

One model handles the full loop: GPT-Realtime-1.5 takes audio in and produces audio out, with no separate transcription and text-to-speech steps. That reduces latency and keeps intonation and conversational cues intact.

### Can GPT-Realtime-1.5 call tools during a live conversation?

Yes. Tool calling works mid-conversation, so a voice agent can look up an order or query a system while the caller is still talking, then respond with the result.

### How does GPT-Realtime-1.5 compare to gpt-realtime-2?

GPT-Realtime-1.5 is the flagship for standard voice agents and customer support. GPT Realtime 2 is the more capable successor, adding configurable reasoning effort for conversations that involve multi-step problem solving.

### How is GPT-Realtime-1.5 priced?

Pricing is listed on this page. Rates reflect the providers routing through AI Gateway and update when providers change list prices.

### Does GPT-Realtime-1.5 support zero data retention through AI Gateway?

Zero Data Retention is not currently available for this model. Zero Data Retention is offered on a per-provider basis. See https://vercel.com/docs/ai-gateway/capabilities/zdr for details.

### What are typical latency characteristics?

This page shows live performance metrics measured across real AI Gateway traffic.

## Links

- [Model page](https://vercel.com/ai-gateway/models/gpt-realtime-1.5)
- [AI Gateway documentation](https://vercel.com/docs/ai-gateway)
- [Provider model documentation](https://developers.openai.com/api/docs/models/gpt-realtime-1.5)


If you encounter incorrect information, can’t find what you need, need a capability Vercel doesn’t currently offer, or have any other feedback to report, please do it by following the instructions at https://vercel.com/feedback.md.