“GPT‑4.1 mini is a big leap in small mannequin efficiency, even beating GPT‑4o in lots of benchmarks. It matches or exceeds GPT‑4o in intelligence evals whereas lowering latency by practically half and lowering price by 83%,” the announcement stated. “For duties that demand low latency, GPT‑4.1 nano is our quickest and least expensive mannequin out there. It delivers distinctive efficiency at a small measurement with its 1 million token context window, and scores 80.1% on MMLU, 50.3% on GPQA, and 9.8% on Aider polyglot coding—even larger than GPT‑4o mini. It’s preferrred for duties like classification or autocompletion.”
These enhancements, OpenAI stated, mixed with primitives such because the Responses API, will enable builders to construct extra helpful and dependable brokers that may carry out advanced duties similar to extracting insights from giant paperwork and resolving buyer requests “with minimal hand-holding.”
OpenAI additionally stated that GPT-4.1 is considerably higher than GPT-4o at duties similar to agentically fixing coding duties, front-end coding, making fewer extraneous edits, following diff codecs reliably, guaranteeing constant instrument utilization, and others.
