Usman Salis
⬤ OpenAI's latest models have climbed to the top of DesignArena.ai's competitive rankings, ending Anthropic's Claude streak at the summit. GPT-5 and GPT-5.1 now lead several ELO-based evaluations, pulling ahead of Claude across creative, design, and multimodal tasks. The jump highlights major gains in OpenAI's newest generation.
⬤ In the first benchmark set, GPT-5.1 (High) scored 1374 ELO, beating Claude 3.7 Sonnet (1317), Claude Opus 4 (1315), and Claude Sonnet 4.5 (1313). Other GPT-5.1 configurations also performed well—Medium hit 1342 and Minimal reached 1343. The "All Categories" panel showed even stronger results: GPT-5.1 (High) reached 1401, well ahead of Claude Opus 4 (1322) and Claude Sonnet 4.5 (1318). The model dominated across UI design, 3D work, data visualization, and image generation.
⬤ GPT-5 also landed strong placements, scoring between 1306 and 1321 depending on the task. While Claude Sonnet 4.5 and Opus variants remain competitive, GPT-5.1 profiles now hold the top spots. The month-over-month shift shows just how fast things are moving as model upgrades and optimization continue to accelerate.
⬤ An updated note confirmed GPT-5.1 is now the #1 overall model on DesignArena.ai, with the observation that "it's crazy what a difference one month of work can make." The rapid improvement since the last evaluation cycle underscores how quickly frontier AI is advancing—and how tight the race has become among leading developers.
Usman Salis
