OpenAI Launches GPT-4.1 With 26% Cost Cut, Responding to DeepSeek’s Pricing Pressure

Key Takeaways

• OpenAI has launched its latest AI model, GPT-4.1.
• GPT-4.1 includes a “nano” version which is OpenAI’s cheapest model yet.
• The budget model competes on price with DeepSeek’s offering.

OpenAI has positioned its latest AI model, GPT-4.1, as a cheaper, more efficient alternative to frontier reasoning models. 

The focus on cost comes as OpenAI faces rising competition from free and open-source alternatives like those offered by DeepSeek.

OpenAI’s Cheapest Model Yet

Compared to its predecessor GPT‑4o, OpenAI said GPT-4.1 is 26% less expensive for median queries. It also costs less than OpenAI’s reasoning models, o1 and o3 mini.

On a per token basis, GPT-4.1 is more than seven times cheaper than o1. Meanwhile, the GPT-4.1 mini costs roughly a third of the price of the o3 mini.

A third, even more lightweight model, GPT-4.1 nano, was described by OpenAI as “our cheapest and fastest model ever,” priced at just $0.1 per million input tokens and $0.4 per million output tokens.

Chatbot Price Wars

The API-only GPT-4.1 is targeted at AI developers rather than end users and hasn’t been integrated into ChatGPT.

But although it is significantly cheaper than previous OpenAI products, the new model competes against an emerging crop of budget alternatives, including DeepSeek’s offering.

The Chinese firm offers some of the most competitive pricing for developers who opt to access DeepSeek models via API rather than deploying locally.

GPT-4.1 nano represents OpenAI’s first serious attempt to cater to the rising demand for low-cost AI services. For API access, it is comparable in price to DeepSeek Chat.

Compared to previous models, GPT-4.1 also introduces additional cost reductions compared to previous models. For example, there are no additional fees for long context requests and an increased discount for cached inputs.

GPT-4.1 vs. DeepSeek Price Comparison

Source: datacamp