OpenAI has launched a new series of multimodal models dubbed GPT-4.1 that represent what the company says is a leap in small model performance, including longer context windows and improvements in coding and instruction following. Geared to developers and available exclusively via API (not through ChatGPT), the 4.1 series comes in three variations: in addition to the flagship GPT‑4.1, GPT‑4.1 mini and GPT‑4.1 nano, OpenAI’s first nano model. Unlike Web-connected models (which have “retrieval-augmented generation,” or RAG) and can access up-to-date information, they are static knowledge models.

Of the 1-million-token prompt window, TechCrunch contextualizes that these models “can take in roughly 750,000 words in one go (longer than ‘War and Peace’),” pointing out that GPT-4.1 “arrives as OpenAI rivals like Google and Anthropic ratchet up efforts,” with Google’s recently released Gemini 2.5 Pro also proffering a 1-million-token context window and flourishing high marks on popular coding benchmarks.

“Anthropic’s Claude 3.7 Sonnet” and Chinese startup DeepSeek’s upgraded DeepSeek-V3 are also part of that mix, TechCrunch adds.

An introductory post offers a “Try in Playground” link for those who have or are willing to sign up for OpenAI accounts. Touting high performance on benchmarks — including the software engineering go-to SWE-bench, the MultiChallenge test of LLM complex conversational ability and the Video MME measurement of multimodal long context understanding — OpenAI also emphasizes “a focus on real-world utility.” Notably, “exceptional performance at a lower cost.”

While improved results for less money is a general goal of every new product release, the AI sector is experiencing downward pressure from less expensive Chinese competition as companies like DeepSeek, Alibaba and Baidu offer what are presented as comparable LLMs at significantly lower costs than their U.S. counterparts.

“GPT-4.1 will cost 26 percent less than its predecessor, while the lightweight nano version becomes OpenAI’s most affordable offering at just 12 cents per million tokens,” VentureBeat writes, quoting OpenAI Chief Product Officer Kevin Weil saying the price reduction comes even as the  new models “are better than GPT-4o on just about every dimension.”

TechCrunch notes that it is “the goal of many tech giants, including OpenAI, to train AI coding models capable of performing complex software engineering tasks,” describing the company’s “grand ambition” as creating an “‘agentic software engineer,’ as CFO Sarah Friar put it during a tech summit in London last month.”

OpenAI has suggested its future models “will be able to program entire apps end-to-end, handling aspects such as quality assurance, bug testing, and documentation writing,” TechCrunch reports, adding that “GPT-4.1 is a step in this direction.”