# GPT-5 pro

The premium tier of the GPT-5 family, providing extended reasoning compute for problems that demand the deepest analysis and highest-quality output OpenAI can deliver.

- **Model ID:** `openai/gpt-5-pro`
- **Type:** chat
- **Providers:** openai
- **Context window:** 400,000
- **Maximum output tokens:** 272,000
- **Pricing:** $15.0/1M input tokens, $120.0/1M output tokens
- **Canonical page:** https://vercel.com/ai-gateway/models/gpt-5-pro

## Supported parameters

Detailed capability metadata has not been reported for this model.

## Example

```ts
import { streamText } from 'ai'

const result = streamText({
  model: 'openai/gpt-5-pro',
  prompt: 'Why is the sky blue?'
})
```

## About

GPT-5 pro became available on October 6, 2025 as the premium tier of the GPT-5 model family on AI Gateway. It applies more compute per request than the standard GPT-5, enabling deeper reasoning, more thorough analysis, and higher-quality output on the most demanding tasks.

The pro tier exists for the subset of problems where standard GPT-5 capability is not enough. Complex research synthesis, thorough code audits, nuanced legal analysis, and multi-factor strategic planning all benefit from the additional reasoning depth GPT-5 pro provides, in exchange for higher cost and potentially longer response times.

GPT-5 pro is typically deployed selectively. Routine traffic flows through GPT-5 or GPT-5 mini, while specific high-value queries are routed to the pro tier. This pattern maintains quality where it matters while keeping overall costs manageable.

## What to consider

GPT-5 pro is for tasks where quality is the only metric that matters. It applies more compute per request than standard GPT-5, producing better results on genuinely hard problems at higher per-request cost.

Most applications should use GPT-5 pro selectively for their hardest queries rather than as a default, pairing it with GPT-5 or GPT-5 mini for routine traffic.

## When to use

### Best For

- **Hard research problems:** Complex analysis that benefits from extended deliberation and thoroughness
- **Critical code review:** Architectural review and security audits where missing an issue has high consequences
- **High-stakes content:** Legal, medical, or financial content where accuracy is paramount
- **Complex multi-step planning:** Strategic decisions and system design that require considering many factors
- **Quality-gated workflows:** Tasks where output is reviewed by humans and quality directly impacts outcomes

### Consider Alternatives When

- **Standard workloads:** GPT-5 handles the vast majority of tasks without the pro tier's premium
- **Cost-sensitive applications:** GPT-5 mini for production traffic at scale
- **Specialized reasoning:** o3-pro for pure mathematical and scientific chain-of-thought problems
- **Speed-critical responses:** Any smaller model when latency matters more than quality ceiling

## Best for

- **Hard research problems:** Complex analysis that benefits from extended deliberation and thoroughness
- **Critical code review:** Architectural review and security audits where missing an issue has high consequences
- **High-stakes content:** Legal, medical, or financial content where accuracy is paramount
- **Complex multi-step planning:** Strategic decisions and system design that require considering many factors
- **Quality-gated workflows:** Tasks where output is reviewed by humans and quality directly impacts outcomes

## Consider alternatives

- **Standard workloads:** GPT-5 handles the vast majority of tasks without the pro tier's premium
- **Cost-sensitive applications:** GPT-5 mini for production traffic at scale
- **Specialized reasoning:** o3-pro for pure mathematical and scientific chain-of-thought problems
- **Speed-critical responses:** Any smaller model when latency matters more than quality ceiling

## Frequently asked questions

### How does GPT-5 pro differ from standard GPT-5?

It applies more compute per request, enabling deeper reasoning and more thorough analysis. Improvements are most apparent on genuinely hard problems; on simple tasks the difference may be minimal.

### When should I use GPT-5 pro versus GPT-5?

Use it selectively for your hardest queries, complex analysis, critical code review, high-stakes content, and tasks where quality directly impacts outcomes. Use standard GPT-5 for routine traffic.

### What context window does GPT-5 pro support?

400K tokens, matching the GPT-5 family.

### Is GPT-5 pro slower than GPT-5?

It may take longer per request due to the additional compute applied. For latency-sensitive applications, consider routing only specific queries to the pro tier.

### How does AI Gateway handle authentication for GPT-5 pro?

AI Gateway accepts a single API key or OIDC token for all requests. You don't embed OpenAI credentials in your application; AI Gateway routes and authenticates on your behalf.

### What are typical latency characteristics?

This page shows live throughput and time-to-first-token metrics measured across real AI Gateway traffic.

## Links

- [Model page](https://vercel.com/ai-gateway/models/gpt-5-pro)
- [AI Gateway documentation](https://vercel.com/docs/ai-gateway)
- [Provider model documentation](https://platform.openai.com/docs/models/gpt-5-pro)
- [Provider pricing](https://platform.openai.com/docs/pricing)