OpenAI announced on Wednesday the rollout of its new GPT-4.1 and GPT-4.1 mini models within ChatGPT, expanding the capability of the popular AI chatbot platform. According to the company, these advanced models are specially optimized for assisting software developers with tasks such as coding and debugging.
An OpenAI representative emphasized that GPT-4.1 significantly surpasses the previously available GPT-4o in both coding proficiency and instruction-following capabilities, while also offering substantially greater speed than the existing “o-series” models designed for general reasoning.
GPT-4.1 will be offered exclusively to subscribers of ChatGPT’s Plus, Pro, and Team tiers. Meanwhile, GPT-4.1 mini, a lighter but still capable version of the model, will be accessible to both paid and free-tier ChatGPT users. With the deployment of these new models, OpenAI has decided to retire GPT-4.0 mini from its platform entirely.
Initially unveiled in April, GPT-4.1 and GPT-4.1 mini had previously only been accessible through OpenAI’s developer API, drawing criticism from some AI researchers due to OpenAI’s decision to not publish a thorough safety report. Researchers had argued this marked a departure from the transparency standards typically practiced for powerful AI model releases. Responding to critics, OpenAI maintained that since GPT-4.1 did not incorporate novel modalities nor surpass the capabilities of the cutting-edge “o3” model algorithmically, it fell outside the classification of a “frontier-model,” which would have warranted strict safety reporting protocols.
Johannes Heidecke, OpenAI’s Head of Safety Systems, reiterated on Wednesday that GPT-4.1 posed a distinctly different set of considerations from the company’s more advanced frontier models, explaining that while GPT-4.1 represented considerable enhancements to coding capabilities, it did not introduce fundamental innovations to warrant extra scrutiny.
Seeking to address transparency concerns, OpenAI also committed on Wednesday to more frequently releasing results from its internal AI safety evaluations. The company has established a new online platform, the Safety Evaluations Hub, to regularly publish these assessments publicly, reflecting a greater openness about its model performance and safety guidelines.
The release of GPT-4.1 coincides with increasing competition among major technology firms to deliver powerful AI coding tools. Google recently updated its own Gemini chatbot to enhance integration with GitHub projects, and OpenAI itself is reportedly close to finalizing a $3 billion acquisition of Windsurf, a leading AI-driven coding utilization tool.
