gpt-4.1-nano
by openai
Pricing
GPT‑4.1 Nano Release by OpenAI
Overview
GPT‑4.1 Nano is the smallest, fastest, and most affordable model in the GPT‑4.1 family, launched via the OpenAI API on April 14, 2025 :contentReference[oaicite:1]{index=1}. It delivers a massive 1 million‑token context window for both text and images, optimized for low-latency, high-throughput tasks such as classification, autocompletion, and lightweight code generation.
- Release Date: April 14, 2025 :contentReference[oaicite:2]{index=2}
- Multimodal Support: Text + Images :contentReference[oaicite:3]{index=3}
- Context Length: Up to 1,000,000 tokens :contentReference[oaicite:4]{index=4}
- Output Limit: Up to ~32K tokens typical :contentReference[oaicite:5]{index=5}
Features
- Ultra‑low latency and lightweight footprint :contentReference[oaicite:6]{index=6}
- Native support for image + text inputs :contentReference[oaicite:7]{index=7}
- Basic code generation (≈ 9.8% on Aider polyglot tasks) :contentReference[oaicite:8]{index=8}
- Extremely cost‑efficient: $0.10 input / $0.40 output per million tokens :contentReference[oaicite:9]{index=9}
Benchmarks
| Task | GPT‑4.1 Nano | GPT‑4.1 Mini | Claude Haiku |
|---|---|---|---|
| MMLU (reasoning) | 80.1 % | ≈ 82 % | 73.8 % |
| GPQA (factual QA) | 50.3 % | ≈ 51.1 % | 45.4 % |
| HumanEval (coding) | 9.8 % | 87.2 % | 64.9 % |
| MMMU (multimodal) | ≈ 53 % | ≈ 59.4 % | ≈ 54.9 % |
Benchmarks sourced from OpenAI blog and community reporting :contentReference[oaicite:10]{index=10}.
Pricing
| Token Type | Cost per 1M tokens |
|---|---|
| Input | $0.10 |
| Output | $0.40 |
Around 90% cheaper than GPT‑4 Turbo :contentReference[oaicite:11]{index=11}.
Use Cases
- Real‑time classification and autocomplete
- Multimodal content parsing (text + images)
- Fast summarization and document tagging
- Lightweight code suggestions and completions
- High‑frequency API applications and embedded agents
Safety & Stability
- Aligns with OpenAI’s instruction-following and safety standards :contentReference[oaicite:12]{index=12}
- No browsing/plugins; API-only access
- Knowledge cutoff: June 2024 :contentReference[oaicite:13]{index=13}
Limitations
- Lower reasoning and coding depth vs. larger GPT‑4.1 variants
- Not available in ChatGPT UI—API access only :contentReference[oaicite:14]{index=14}
- No audio/video support yet
License
Commercial use is permitted under OpenAI’s API terms. GPT‑4.1 Nano is optimized for integration into real‑time tools, mobile apps, and scalable embedded LLM environments.