Anthropic’s Claude Opus 5 has claimed the top spot on the Fullstack Code Arena leaderboard with a score of 1,699 points, outperforming OpenAI’s GPT-5.6 Sol which scored approximately 1,638 points. This new benchmark, launched in August 2026 on Arena.ai, evaluates end-to-end web development capabilities of AI models, including multi-step reasoning, tool use, database integration, and API orchestration. The model was also released on July 24, 2026, priced at $5 per million input tokens and $25 per million output tokens. Moonshot AI’s Kimi K3 also emerged as a competitive presence, though its exact score was not specified in the latest rankings.
Source: Read the original article
