OpenAI has released special AI models for IT professionals.

OpenAI has announced a new family of models aimed at software development tasks: GPT-4.1, GPT-4.1 mini, and GPT-4.1 nano. All variants are specialized in programming, support a 1 million-token context window (approximately 750 million words), and a 32,000-token output window. The models are currently accessible exclusively via API. Pricing: - GPT-4.1: $2 per 1 million input tokens and $8 per 1 million output tokens.

- GPT-4.1 nano: $0.10 per 1 million input tokens and $0.40 per 1 million output tokens.

According to OpenAI, GPT-4.1 is optimized for front-end development, format compliance, and tool usage. On the SWE-bench Verified benchmark, the model achieves a performance range of 52–54.6%, outperforming GPT-4o and GPT-4o mini, while remaining slightly below Gemini 2.5 Pro (63.8%) and Claude 3.7 Sonnet (62.3%).

Model reliability reportedly decreases with larger context usage — from 84% at 8,000 tokens to 50% at 1 million tokens.

OpenAI also confirmed the upcoming removal of GPT-4 from ChatGPT as of April 30, designating GPT-4o as its replacement. GPT-4.5 will be deprecated in the API on July 14, citing GPT-4.1’s improved or equivalent performance at lower cost and latency.

GPT-4o, which now serves as the default model in ChatGPT, recently gained image generation capabilities. Due to high demand, OpenAI temporarily restricted access for free-tier users to manage GPU load.

Additionally, OpenAI CEO Sam Altman announced on April 4 that the release of GPT-5 has been postponed and will not take place in May, as previously expected. Full versions of the o3 and o4 mini reasoning models are anticipated to launch soon, with references discovered in a recent ChatGPT web update by engineer Tibor Blaho.