Overview
GPT-5 nano is the smallest, fastest, and lowest-cost model in OpenAI's GPT-5 family, released on August 7, 2025 alongside GPT-5 and GPT-5 mini. It is an API-only model aimed at high-volume, latency-sensitive work — OpenAI positions it for tasks like summarization and classification where speed and cost matter more than maximum reasoning depth.
Despite its size, GPT-5 nano carries the generation's full 400,000-token context window and can produce up to 128,000 output tokens. It accepts both text and image input and returns text. Like the rest of the GPT-5 line it is a reasoning model with a configurable reasoning effort, so developers can dial thinking up for harder prompts or down to keep responses near-instant and cheap.
Pricing is the headline: at $0.05 per million input tokens and $0.40 per million output tokens, GPT-5 nano is OpenAI's cheapest GPT-5 model, with cached input at $0.005 per million tokens cutting the cost of repeated context further. It is offered via the OpenAI API as model id gpt-5-nano (snapshot gpt-5-nano-2025-08-07), and its knowledge cutoff is May 2024. It is the first entry in the GPT Nano line, succeeded by GPT-5.4 nano.
| Released | 2025-08-07 |
|---|---|
| License | Proprietary |
| Weights | API only |
| Parameters | Undisclosed |
| Context | 400K |
| Max output | 128K |
| Architecture | Undisclosed |
| Knowledge cutoff | May 2024 |
| Modalities | Text, Vision |
| Status | Generally available |
Pricing
| Input | $0.05 / 1M tokens |
|---|---|
| Cached input | $0.005 / 1M tokens |
| Output | $0.40 / 1M tokens |
Strengths
- Lowest-cost GPT-5 model — $0.05 input / $0.40 output per 1M tokens, with cached input at $0.005
- Very fast, low-latency responses suited to real-time and high-volume traffic
- Full 400K-token context window despite being the smallest GPT-5 variant
- Text + image input, so it can read screenshots and documents on a budget
- Configurable reasoning effort to trade speed against accuracy per request
Best for
- Reach for it for high-throughput classification, tagging, and routing where per-call cost dominates.
- Reach for it for summarization of large volumes of text at the lowest price point.
- Reach for it as the cheap tier in a model-routing setup, escalating only hard prompts to GPT-5 mini or GPT-5.
How to access
| Provider | Model ID |
|---|---|
| OpenAI API ↗ | gpt-5-nano |
| OpenRouter ↗ | openai/gpt-5-nano |
GPT Nano — every version
The full lineage of the GPT Nano line, newest first. Every version has its own page — click any to compare specs, benchmarks and pricing.
| Version | Released | Context | License |
|---|---|---|---|
| GPT-5.4 nanocurrent | 2026-03-17 | — | Proprietary |
| GPT-5 nano | 2025-08-07 | — | Proprietary |
FAQ
What is GPT-5 nano?
GPT-5 nano is the smallest, fastest, and cheapest model in OpenAI's GPT-5 family, released on August 7, 2025. It is an API-only reasoning model built for high-volume, latency-sensitive tasks such as summarization and classification, where speed and cost matter more than maximum reasoning depth.
How much does GPT-5 nano cost?
On the OpenAI API, GPT-5 nano costs $0.05 per million input tokens and $0.40 per million output tokens, with cached input priced at $0.005 per million tokens. That makes it OpenAI's cheapest GPT-5 model.
What is the context window of GPT-5 nano?
GPT-5 nano has a 400,000-token context window and can generate up to 128,000 output tokens — the same context limit as the larger GPT-5 and GPT-5 mini models.
Does GPT-5 nano support image input?
Yes. GPT-5 nano accepts both text and image input and returns text. Its knowledge cutoff is May 2024, and it is served via the OpenAI API as model id gpt-5-nano (snapshot gpt-5-nano-2025-08-07).