# GPT-5 nano

The smallest and fastest GPT-5 family model, built for classification, routing, and high-volume lightweight inference at the lowest cost in the lineup.

- **Model ID:** `openai/gpt-5-nano`
- **Type:** chat
- **Providers:** azure, openai
- **Context window:** 400,000
- **Maximum output tokens:** 128,000
- **Pricing:** $0.05/1M input tokens, $0.40/1M output tokens
- **Canonical page:** https://vercel.com/ai-gateway/models/gpt-5-nano

## 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-nano',
  prompt: 'Why is the sky blue?'
})
```

## About

GPT-5 nano was released on August 7, 2025 as the entry-level tier of the GPT-5 model family. It's optimized for the highest throughput and lowest latency in the family, targeting workloads where speed and cost matter more than reasoning depth.

Despite being the smallest GPT-5 variant, GPT-5 nano benefits from the family's architectural improvements. It handles classification, routing, extraction, and simple generation tasks with quality that reflects the generational leap from GPT-4.1 nano. The context window of 400K tokens is notable for a model at this tier, enabling it to process long inputs even when outputs remain short.

The model is designed to serve as a building block in larger systems: classifying incoming requests, routing them to appropriate handlers, extracting key fields from documents, and providing instant responses for simple queries, all at a cost that makes per-request inference viable for the highest-traffic applications.

## What to consider

GPT-5 nano prioritizes throughput and latency over reasoning depth. It's the right choice when you need fast answers to simple questions at minimal cost.

At its price point, GPT-5 nano is practical as a classifier, router, or preprocessor that runs on every request, deciding which downstream model or action to invoke.

## When to use

### Best For

- **Real-time classification:** Sentiment analysis, intent detection, and topic labeling at high request volume
- **Routing and triage:** Deciding which model or workflow handles each incoming request
- **Autocomplete and suggestions:** Sub-second inline suggestions in editors and search interfaces
- **Lightweight extraction:** Pulling specific fields from structured or semi-structured text
- **Cost-sensitive batch processing:** Millions of simple inferences at minimal aggregate cost

### Consider Alternatives When

- **Complex reasoning needed:** GPT-5 mini or GPT-5 for tasks requiring multi-step analysis
- **Code generation:** Codex mini or GPT-5 codex for coding-specific tasks
- **Deep deliberation:** o3 or o4-mini for problems that benefit from chain-of-thought reasoning
- **Rich multimodal analysis:** Full GPT-5 for complex vision and document understanding tasks

## Best for

- **Real-time classification:** Sentiment analysis, intent detection, and topic labeling at high request volume
- **Routing and triage:** Deciding which model or workflow handles each incoming request
- **Autocomplete and suggestions:** Sub-second inline suggestions in editors and search interfaces
- **Lightweight extraction:** Pulling specific fields from structured or semi-structured text
- **Cost-sensitive batch processing:** Millions of simple inferences at minimal aggregate cost

## Consider alternatives

- **Complex reasoning needed:** GPT-5 mini or GPT-5 for tasks requiring multi-step analysis
- **Code generation:** Codex mini or GPT-5 codex for coding-specific tasks
- **Deep deliberation:** o3 or o4-mini for problems that benefit from chain-of-thought reasoning
- **Rich multimodal analysis:** Full GPT-5 for complex vision and document understanding tasks

## Frequently asked questions

### What tasks is GPT-5 nano designed for?

Classification, routing, autocomplete, lightweight extraction, and any high-volume workload where speed and cost outweigh the need for deep reasoning.

### How does GPT-5 nano compare to GPT-4.1 nano?

GPT-5 nano is the next generation, inheriting GPT-5 family improvements in quality and instruction following while maintaining the speed and cost profile expected of a nano-tier model.

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

400K tokens, which is substantial for a model at this price and speed tier.

### Can GPT-5 nano handle long documents?

It can read long inputs within its window of 400K tokens, but it's optimized for short outputs. For detailed analysis of long documents, consider GPT-5 mini or GPT-5.

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

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-nano)
- [AI Gateway documentation](https://vercel.com/docs/ai-gateway)
- [Provider model documentation](https://learn.microsoft.com/en-us/azure/ai-foundry/foundry-models/concepts/models-sold-directly-by-azure?pivots=azure-openai&tabs=global-standard-aoai%2Cstandard-chat-completions%2Cglobal-standard#gpt-5.1)
- [Provider pricing](https://platform.openai.com/docs/pricing)