OpenAI recently introduced the GPT-4.1 model family, which includes variants such as GPT-4.1 mini and GPT-4.1 nano, each designed to excel in coding and instruction following. These multimodal models, available via OpenAI’s API rather than ChatGPT, feature a context window of 1 million tokens, allowing them to process a substantial amount of text, surpassing the length of “War and Peace.” As competitors like Google and Anthropic enhance their programming models, OpenAI aims to develop AI systems capable of executing advanced software engineering tasks. The company’s ambition, as expressed by CFO Sarah Friar at a recent summit, is to create an “agentic software engineer” that can manage complete application development processes, including quality assurance and documentation.
GPT-4.1 is a step towards this ambition. It has been optimized based on developer feedback to improve critical aspects of software engineering tasks, such as frontend coding and consistent tool application. OpenAI asserts that GPT-4.1 outperforms its predecessors on coding benchmarks and that its mini and nano versions, while faster, sacrifice some accuracy. Pricing for GPT-4.1 is set at $2 per million input tokens and $8 per million output tokens, with the mini and nano versions priced lower.
OpenAI’s evaluations indicate GPT-4.1 generates more tokens at once than previous models, achieving notable performance in benchmarks such as SWE-bench Verified. However, it still lags behind other AI competitors like Google’s Gemini 2.5 Pro and Anthropic’s Claude 3.7 Sonnet on certain metrics. The model also demonstrated a 72% accuracy in the Video-MME evaluation for video content comprehension. Despite these advancements, like other leading models, it struggles with routine coding tasks, which can introduce security vulnerabilities instead of resolving them. OpenAI acknowledges that GPT-4.1’s reliability diminishes with an increase in input tokens, underscoring the need for precise prompts to elicit good performance from the model.
The ainewsarticles.com article you just read is a brief synopsis; the original article can be found here: Read the Full Article…
