Code Arena, available at arena.ai, has launched an Image-to-WebDev benchmark ranking large language models on converting UI design screenshots into functional web code. Opus 5 (Max) tops the rankings, followed by GPT-5.6 Sol in second, with Grok-4.5, Kimi K3, Muse Spark, GPT-5.6 Terra, and Luna completing the upper tier. The benchmark launched on April 15, 2026, and has been adding new models continuously since. The platform evaluates agentic coding workflows, involving complex multi-step reasoning and tool use, to produce working HTML and React code from images or mockups. The leading model, claude-opus-5-max, scored 1703 points on related WebDev leaderboards as of late July 2026, reflecting performance across UI cloning and iterative coding tasks. Code Arena has no direct connection to cryptocurrency or blockchain, being a pure AI evaluation platform. The top seven models span at least five AI providers, including Anthropic, OpenAI, and xAI, plus newer entrants like Kimi K3 and Muse Spark. GPT-5.6 appears in three variants—Sol, Terra, and Luna—suggesting providers are fine-tuning distinct versions for different coding strengths. The benchmark explicitly tests React code generation, and while the 1703-point score establishes a high-water mark, Code Arena has added new Claude variants and other models through July 2026, keeping rankings fluid.