Claude, the most efficient AI for work according to OpenAI

OpenAI’s study shows Claude Opus 4.1 outperforms GPT‑5 on professional tasks, delivering higher productivity and superior document styling.

According to OpenAI, Claude is the most efficient AI for work

OpenAI’s internal study puts Anthropic’s Claude Opus 4.1 ahead of competing large‑language models when it comes to real‑world professional tasks. The benchmark, called “GDPval,” evaluates the economic value generated by AI responses across scenarios such as drafting e‑mail replies, formatting documents, optimizing tables, and auditing pricing inconsistencies. Claude achieved a 43.56 % win rate with a 3.99 % tie rate, yielding a combined score of 47.55 %, comfortably surpassing GPT‑5 high’s 35.48 % win rate. The advantage is most pronounced in aesthetics and presentation, where Claude excels at slide layout, report formatting, and visual consistency.

The research was conducted by OpenAI’s Economic Research team together with NBER economist David Deming. While Claude leads on polish and readability, OpenAI notes that GPT‑5 retains an edge on factual accuracy and domain‑specific knowledge retrieval. This split suggests a strategic choice: use Claude for tasks that demand clean, well‑styled outputs, and rely on GPT‑5 for deep analytical or data‑intensive work.

For businesses, the findings carry tangible implications. Deploying Claude for content creation, marketing copy, and internal communications can shave minutes off the editing cycle and improve brand consistency. Meanwhile, technical teams will likely continue to favor GPT‑5 for code generation, complex problem solving, and precise research queries. Balancing the two models lets organizations optimize cost while maximizing productivity.

OpenAI’s report also highlights the rapid pace of model improvement. The jump from GPT‑4o to GPT‑5 doubled accuracy metrics within a year, and Claude Opus 4.1 already shows a marked performance uplift over earlier Anthropic releases. This trend pushes AI providers toward economic‑focused evaluations rather than purely technical benchmarks, aligning with professional users who demand measurable efficiency gains.

In summary, OpenAI’s claim that Claude is the most efficient AI for work rests on recent, data‑driven evidence. Decision‑makers are encouraged to pilot both Claude and OpenAI models within their own workflows, selecting the tool that best matches each specific need. The competitive dynamics between Claude and OpenAI’s lineup promise continued innovation and increasingly powerful assistance for the modern workplace.

Sources :

• OpenAI tested GPT‑5, Claude, and Gemini on real‑world tasks – the results were surprising | ZDNET (2025)

• OpenAI says GPT‑5 and Claude AI are close to matching human experts in key jobs – India Today (2025)

• Claude just beat GPT‑5, Gemini, and Grok in real‑world job tasks, according to OpenAI’s own study | TechRadar (2025)

Related Stories