GPT-4.1 Nano

openaiopenai/gpt-4.1-nano

Fastest, most cost-efficient version of GPT-4.1. Excels at instruction following and tool calling with a 1M token context window and minimal latency.

GPT-4.1 announcement ↗API docs ↗

GPT-4.1 Nano is OpenAI's fastest and most cost-efficient model, released alongside GPT-4.1 and GPT-4.1 Mini in April 2025. As the smallest member of the GPT-4.1 family, it is purpose-built for tasks where latency and per-token cost dominate — classification, autocompletion, entity extraction, and high-throughput data pipelines — while still supporting a 1-million-token context window and vision input.

Despite its compact size, GPT-4.1 Nano scores 80.1% on MMLU, 50.3% on GPQA, and 9.8% on Aider's polyglot coding benchmark — surpassing GPT-4o Mini on every metric. For router operators, it is the go-to model when you need a capable instruction follower at the absolute lowest cost on the latency curve, especially in agentic loops where each step must be fast and cheap.

Best for

• Real-time classification and intent routing where sub-200ms latency is required
• Autocompletion and inline suggestions in IDEs or search boxes at massive scale
• Structured data extraction and entity recognition from documents using the 1M context window
• High-volume agentic loops where each step must be fast and cheap — e.g. tool-selection routing, log triage, ticket tagging

Reach for something else if

• Complex multi-step reasoning or hard math — route to o3, o4-mini, or GPT-5 instead
• Creative long-form writing that demands nuance and personality — GPT-4.1 or GPT-5 are better suited
• Vision-intensive multimodal analysis — GPT-4.1 or GPT-5 deliver significantly better image comprehension
• Large code generation or agentic coding tasks — GPT-4.1 Mini offers a better intelligence-to-cost ratio for these

Context Length

1.0M

Max Output

33K

Input Priceper 1M tokens

$0.108/ 1M tokens

Output Priceper 1M tokens

$0.432/ 1M tokens

Modalities

textimage→text

Capabilities

Vision

Pricing Breakdown

Type	Rate
Input	$0.108 / 1M tokens
Output	$0.432 / 1M tokens

Supported Parameters

temperaturemax_tokenstop_ptoolstool_choiceresponse_formatstop

Specifications

Release date	2025-04-14openai.com ↗	verified
Knowledge cutoff	June 2024openai.com ↗	verified
Training cutoff	June 2024openai.com ↗	verified
License	Proprietary — OpenAI API onlyopenai.com ↗	verified
Model family position	Smallest and fastest in GPT-4.1 family (Nano < Mini < 4.1)openai.com ↗	verified

Benchmarks

Benchmark	Score	Source
MMLU Reported in GPT-4.1 announcement; higher than GPT-4o Mini	80.1%%	openai.com ↗
GPQA Graduate-level Q&A benchmark; notable for a nano-class model	50.3%%	openai.com ↗
Aider Polyglot Multi-language coding benchmark; higher than GPT-4o Mini	9.8%%	openai.com ↗

API Usage Examples

Use the global api.therouter.ai endpoint shown below for new integrations; the legacy China accelerated endpoint is retired.

cURL

curl https://api.therouter.ai/v1/chat/completions   -H "Content-Type: application/json"   -H "Authorization: Bearer $THE_ROUTER_API_KEY"   -d '{
    "model": "openai/gpt-4.1-nano",
    "messages": [
      {"role": "user", "content": "Summarize the key points from this input."}
    ]
  }'

API guide

Full API reference →

Chat completion

cURL

curl https://api.therouter.ai/v1/chat/completions \
  -H "Authorization: Bearer $THEROUTER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"openai/gpt-4.1-nano","messages":[{"role":"user","content":"Hello"}]}'

More from openai

openai/gpt-4.1Full-sized GPT-4.1 — highest benchmark scores in the family openai/gpt-4.1-miniMid-tier GPT-4.1 — balances intelligence and cost between 4.1 and Nano openai/gpt-4o-miniPrevious-gen mini model — GPT-4.1 Nano exceeds it on MMLU, GPQA, and Aider

Similar models

Cross-provider sibling models

google/gemini-2.5-flash

Google's cost-efficient model with 1M+ context — similar speed tier

anthropic/claude-haiku-4.5

Anthropic's fastest model — comparable latency with strong tool use

meta/llama-4-maverick

Open-weight alternative for self-hosted low-latency inference

News & changes

2025-04-14

GPT-4.1 family launched with first-ever Nano model

OpenAI releases GPT-4.1, GPT-4.1 Mini, and GPT-4.1 Nano with 1M-token context. Nano is OpenAI's first model explicitly targeting the ultra-low-latency, ultra-low-cost tier — scoring 80.1% on MMLU and exceeding GPT-4o Mini on coding benchmarks, while positioned as the fastest model on the API latency curve.

re-authored by TheRouteropenai.com ↗

Frequently asked

How does GPT-4.1 Nano compare to GPT-4o Mini?

GPT-4.1 Nano outperforms GPT-4o Mini across MMLU (80.1% vs lower), GPQA (50.3%), and Aider polyglot coding (9.8%). It also supports a 1-million-token context window (vs 128K for GPT-4o Mini) at a lower per-token cost. It is positioned as the replacement for GPT-4o Mini in the lowest-cost API tier.

openai.com ↗

What is GPT-4.1 Nano best used for?

OpenAI recommends GPT-4.1 Nano for tasks that demand low latency above all — classification, autocompletion, entity extraction, and high-throughput data pipelines. It is also a strong choice as the fast-routing step in agentic systems: use Nano to classify intent or select tools, then route complex work to GPT-4.1 or o3.

openai.com ↗

Can GPT-4.1 Nano handle vision tasks?

Yes — GPT-4.1 Nano accepts image inputs alongside text, making it suitable for cheap, high-throughput image classification or OCR pre-processing. For detailed image understanding or complex visual reasoning, GPT-4.1 or GPT-5 will produce better results.

How do I call GPT-4.1 Nano via TheRouter?

Use the standard OpenAI SDK with baseURL set to https://api.therouter.ai/v1 and model set to "openai/gpt-4.1-nano". TheRouter is fully OpenAI-compatible — no SDK changes needed beyond the base URL and model name. See the API guide section above for curl, TypeScript, and Python examples.

Fact ledger — every claim on this page traces here

source	URL	retrieved
Release date	openai.com ↗	2026-05-28	verified
Knowledge cutoff	openai.com ↗	2026-05-28	verified
Training cutoff	openai.com ↗	2026-05-28	verified
License	openai.com ↗	2026-05-28	verified
Model family position	openai.com ↗	2026-05-28	verified
MMLU	openai.com ↗	2026-05-28	verified
GPQA	openai.com ↗	2026-05-28	verified
Aider Polyglot	openai.com ↗	2026-05-28	verified
GPT-4.1 family launched with first-ever Nano model	openai.com ↗	2026-05-28	verified
How does GPT-4.1 Nano compare to GPT-4o Mini?	openai.com ↗	2026-05-28	to verify
What is GPT-4.1 Nano best used for?	openai.com ↗	2026-05-28	to verify