GPT-5 nano API Pricing: OpenAI's Cheapest Model at $0.05/M Tokens

Q: How much does GPT-5 nano cost?

GPT-5 nano costs $0.05 per million input tokens and $0.40 per million output tokens. This makes it OpenAI's cheapest model — 50x cheaper than GPT-5.4 on input and 37x cheaper on output.

Q: What is the context window of GPT-5 nano?

GPT-5 nano has a 128K token context window. This is sufficient for most chat, classification, and extraction tasks, though smaller than the 1M windows offered by Qwen 3.7 Flash or Gemini 2.5 Flash-Lite.

Q: Is GPT-5 nano cheaper than Qwen 3.7 Flash?

No. Qwen 3.7 Flash ($0.03/$0.13) is cheaper than GPT-5 nano ($0.05/$0.40) on both input (1.7x) and output (3x). However, GPT-5 nano runs on OpenAI's API with guaranteed uptime and SLA, which may matter for production workloads.

Q: What can GPT-5 nano be used for?

GPT-5 nano is ideal for classification, data extraction, content moderation, simple Q&A, and translation — any high-volume task where speed and cost matter more than complex reasoning. For agentic workflows or nuanced writing, upgrade to GPT-5.4 or Claude Sonnet 5.

GPT-5 nano is OpenAI's lowest-cost model — 50x cheaper than GPT-5.4 on input, with the same API and infrastructure reliability.

Updated Aug 3, 2026 · 93 models tracked across 11 providers

TL;DR

Price: $0.05/M input, $0.40/M output — OpenAI's cheapest model by a wide margin
Context: 128K tokens — sufficient for most chat, classification, and extraction tasks
Vision: No — text-only input. For cheapest vision, use Qwen 3.7 Flash ($0.03/$0.13) or Gemini 2.5 Flash-Lite ($0.10/$0.40)
Provider: OpenAI API (platform.openai.com)
Best for: High-volume classification, data extraction, content moderation, simple Q&A, translation
Trade-off: Smaller context (128K vs 1M) and less capable than GPT-5.4 for complex reasoning

GPT-5 nano Pricing Breakdown

At $0.05 per million input tokens and $0.40 per million output tokens, GPT-5 nano is 50x cheaper on input than GPT-5.4 ($2.50/$15). Here's how the costs scale:

Monthly Volume	Input Cost	Output Cost	Total Cost
1M tokens	$0.05	$0.40	$0.45
10M tokens	$0.50	$4.00	$4.50
100M tokens	$5.00	$40.00	$45.00
1B tokens	$50.00	$400.00	$450.00

At 100M tokens/month, GPT-5 nano costs $45. The same volume on GPT-5.4 would cost $1,275, and on Claude Sonnet 5 it would cost $900+.

How GPT-5 nano Compares to Other Budget Models

Model	Input $/M	Output $/M	Context	Vision	Provider
Qwen 3.7 Flash	$0.03	$0.13	1M	✅	Alibaba
GPT-5 nano	$0.05	$0.40	128K	❌	OpenAI
GPT-oss 20B	$0.08	$0.35	128K	❌	OpenAI (self-host)
Gemini 2.5 Flash-Lite	$0.10	$0.40	1M	✅	Google
Ministral 3 3B	$0.10	$0.10	128K	❌	Mistral
DeepSeek V4 Flash	$0.14	$0.28	1M	❌	DeepSeek
Mistral Small 4	$0.15	$0.60	128K	❌	Mistral

GPT-5 nano is the cheapest model available on OpenAI's API. Qwen 3.7 Flash is cheaper overall, but GPT-5 nano offers OpenAI's infrastructure, SLA, and ecosystem — including function calling, structured outputs, and batch API support.

GPT-5 nano vs GPT-5.4 nano: Is the Upgrade Worth It?

GPT-5.4 nano ($0.20/$1.25) is OpenAI's next tier up. Here's what you get for the 4x price increase:

Feature	GPT-5 nano	GPT-5.4 nano
Input price	$0.05/M	$0.20/M
Output price	$0.40/M	$1.25/M
Context window	128K	128K
Reasoning	Basic	Improved
Function calling	✅	✅
Structured outputs	✅	✅
Best for	High-volume, simple tasks	General-purpose, moderate complexity

Rule of thumb: Use GPT-5 nano for classification, extraction, and simple Q&A. Upgrade to GPT-5.4 nano when you need better reasoning, multi-step logic, or more nuanced outputs.

When to Use GPT-5 nano

📊 Classification

Sentiment analysis, intent detection, content categorization. High volume, low complexity — ideal for the cheapest OpenAI model.

🔍 Data Extraction

Parse invoices, extract fields from documents, structured data from unstructured text. Works well with function calling for structured output.

🛡️ Content Moderation

Flag inappropriate content, spam detection, policy violations. High throughput at minimal cost.

💬 Simple Q&A

FAQ bots, knowledge base lookups, internal helpdesk. Works well when answers are in the provided context.

🌐 Translation

Basic translation for internal tools or draft content. For customer-facing translations, consider a quality check pass.

📝 Summarization

Short-form summarization of documents, emails, or support tickets. 128K context handles most documents.

When NOT to Use GPT-5 nano

GPT-5 nano is optimized for speed and cost, not complexity. Upgrade when you need:

Complex reasoning: Multi-step logic, math, or code generation → use GPT-5.4 ($2.50/$15) or Claude Sonnet 5 ($2/$10)
Nuanced writing: Creative content, marketing copy, or tone-sensitive communication → use Claude Opus 5 ($5/$25)
Agentic workflows: Tool use, multi-step planning, or autonomous decision-making → use GPT-5.4 or Claude Sonnet 5
Large context: Documents over 128K tokens → use Qwen 3.7 Flash (1M, $0.03/$0.13) or Gemini 2.5 Flash-Lite (1M, $0.10/$0.40)
Vision: Image inputs → use Qwen 3.7 Flash ($0.03/$0.13) or Gemini 2.5 Flash-Lite ($0.10/$0.40)

Real-World Cost Scenario

Let's say you're building a content moderation pipeline that processes 1,000,000 items per month, with an average of 500 input tokens and 200 output tokens per item:

Model	Monthly Input	Monthly Output	Total
Qwen 3.7 Flash	$15	$26	$41
GPT-5 nano	$25	$80	$105
Gemini 2.5 Flash-Lite	$50	$80	$130
GPT-5.4 nano	$100	$250	$350
Claude Haiku 4.5	$500	$1,000	$1,500

GPT-5 nano costs $105/month for 1M moderation items — 3.3x cheaper than GPT-5.4 nano and 14x cheaper than Claude Haiku 4.5.

Frequently Asked Questions

How much does GPT-5 nano cost?

$0.05 per million input tokens and $0.40 per million output tokens. This is OpenAI's cheapest model — 50x cheaper than GPT-5.4 on input and 37x cheaper on output.

What is the context window of GPT-5 nano?

128K tokens. This is sufficient for most chat, classification, and extraction tasks, though smaller than the 1M windows offered by Qwen 3.7 Flash or Gemini 2.5 Flash-Lite.

Is GPT-5 nano cheaper than Qwen 3.7 Flash?

No. Qwen 3.7 Flash ($0.03/$0.13) is cheaper on both input (1.7x) and output (3x). However, GPT-5 nano runs on OpenAI's API with guaranteed uptime, SLA, and ecosystem features like function calling and structured outputs.

Is GPT-5 nano being deprecated?

No. GPT-5 nano is a current model with no announced deprecation date. It is OpenAI's recommended budget-tier model for high-volume, low-complexity tasks.

What can GPT-5 nano be used for?

Classification, data extraction, content moderation, simple Q&A, translation, and summarization — any high-volume task where speed and cost matter more than complex reasoning. For agentic workflows or nuanced writing, upgrade to GPT-5.4 or Claude Sonnet 5.

Calculate Your GPT-5 nano Costs

Enter your token counts and see exactly what you'd pay. Compare against 93 other models instantly.

Open Calculator → See Full Ranking