GPT-4.1 nano

GPT-4.1 Nano is the entry-level variant of OpenAI's GPT-4.1 family, launched alongside GPT-4.1 and GPT-4.1 mini as the company's first-ever nano-tier model. The family shares an emphasis on long-context comprehension, supporting up to one million tokens, and was trained with a refreshed June 2024 knowledge cutoff. Within this lineup, GPT-4.1 Nano is positioned as the fastest and most cost-efficient option, targeting low-latency workloads such as classification, extraction, routing, tagging, and code or text autocompletion where large volumes of short requests make per-token economics critical.

Although GPT-4.1 Nano is the smallest member of its family, it retains the same one-million-token context window and multimodal input handling as its siblings, allowing it to ingest long documents, PDFs, and images alongside text. Third-party benchmarks reported on routing and documentation sites score it at 80.1 percent on MMLU, 50.3 percent on GPQA, and 9.8 percent on the Aider polyglot coding evaluation, with the coding and instruction-following figures exceeding those of GPT-4o mini. These characteristics make it a practical choice for embedding it into tiered retrieval or routing architectures where a lightweight, cheap model can handle the bulk of traffic while larger GPT-4.1 variants are reserved for harder reasoning tasks.

Abacusgpt-4.1-nanogpt-nano

Open provider page Provider docs

Quick Info

Provider: Abacus
Model key: gpt-4.1-nano
Release date: Apr 14, 2025
Last updated: Apr 14, 2025
Knowledge cutoff: 2024-04
Input modalities
Output modalities
Capabilities

Cost

Input token cost: $0.10
Output token cost: $0.40

Limits

Output tokens: 32,768 tokens
Context window: 1,047,576 tokens

Transparent token rates

Compare GPT-4.1 nano pricing

Rates are shown per one million tokens. Combined means one million input plus one million output tokens.

Browse this family

Latest news about GPT-4.1 nano

Abacus

Official sourceAnnouncement

OpenAI launched GPT-4.1, GPT-4.1 mini, and GPT-4.1 nano in the API on April 14, 2025, making nano the fastest and cheapest model in the family with a 1 million token context window. The launch post notes nano scores 80.1% on MMLU, 50.3% on GPQA, and 9.8% on Aider polyglot coding, exceeding GPT-4o mini. It is positioned for low-latency use cases such as classification and autocompletion. The GPT-4.1 family ships with a refreshed June 2024 knowledge cutoff and prioritizes coding, instruction following, and long-context comprehension. OpenAI highlights that GPT-4.1 mini matches or surpasses GPT-4o on intelligence benchmarks while roughly halving latency and cutting cost by 83%, framing nano as the cost-efficient endpoint of the same lineup. GPT-4.1 itself scores 54.6% on SWE-bench Verified, 38.3% on Scale MultiChallenge, and 72.0% long-no-subtitles on Video-MME.

Abacus

Official sourceDocumentation

OpenAI's developer documentation lists GPT-4.1 nano as Deprecated and recommends starting with GPT-5 nano for more complex tasks. The model is described as the fastest, most cost-efficient version of GPT-4.1, excelling at instruction following and tool calling with low latency and no reasoning step. The docs specify a 1,047,576-token context window, 32,768 max output tokens, text and image input with text output, audio and video not supported, and a June 1, 2024 knowledge cutoff. Pricing is set at $0.10 per 1M input tokens, $0.025 cached input, and $0.40 per 1M output, with Batch API discounts available. Supported endpoints include Chat Completions, Responses, Realtime, Batch, and Fine-tuning under the v1 API.

Azure Cognitive Services

CoverageBenchmark

GPT-4.1 Nano, developed by OpenAI, is the smallest member of the GPT-4.1 family, supporting a 1M-token context window, a 32.8K-token output limit, and multimodal inputs (image, text, file) with text outputs and the GPT tokenizer. According to the page, it was released on April 14, 2025, with a knowledge cutoff of June Pricing on the page is listed at $0.10 per million input tokens and $0.40 per million output tokens, with a $0.02 per million cache-read cost and $0.01 per web search operation. Benchmark figures include a Coding Index of 11.1 (Artificial Analysis) and a Best Design Arena score of 993 Elo (111 in Game Dev with a 49.6%