News
Claude Opus 5
2 min read
Source: Telegram AI频道
Claude Opus 5.5 Max strongly topped the list of intelligent agents, and domestic multi-models were firmly in the middle and upper reaches. The new Arena AI large model weekly list was officially announced. Major manufacturers launched a new round of strength competition in multiple sub-lists such as comprehensive, code, intelligent agent and multimedia. In this issue of the list, the new model launched by Anthropic performs extremely well, while the domestic large models maintain a solid echelon performance in various segments. In the smart agent list that has attracted much attention in the industry, Anthropic’s Claude Opus5.5 (Max) showed an overwhelming advantage as a new entry into the list and won the crown in one fell swoop. Its task confirmation success rate reached 17.29%, and the tool hallucination rate was controlled at an ultra-low level of 0.37%. At the same time, domestic large models are making steady progress in the smart body track, and many products such as DeepSeek V4.1Flash (Max), Dark Side of the Moon Kimi K3 (Max), Tencent Hy4preview and Xiaomi MiMo V2.6Pro are among the forefront, maintaining a stable position in the mid-to-upper echelon. The competition at the top of the comprehensive list and code list is also fierce. In the comprehensive list, Google's Gemini-4-argon-high continues to occupy the top spot, followed by Anthropic's various main models, taking many of the top four seats; in terms of domestic models, Dark Side of the Moon kimi-k3-max ranks 16th, Alibaba qwen3.8-max, Xiaomi mimo-v2.6-pro and Zhipu glm-5.3-max and others have also maintained stable rankings in the list. In the code list, Gemini-4-argon-high continues to lead the pack, while Claude Opus5.5 (High) jumps sharply into the top five. The code processing capabilities of domestic models such as Dark Side of the Moon and Xiaomi also continue to occupy a solid position at the forefront of the industry. In front-end development and multi-modal graphics, video and other segmented tracks, the technology iterations of various manufacturers have not slowed down. On the front-end development list, Claude Opus5.5Max continues to lead with its excellent interaction and rendering capabilities. Xiaomi's new mimo-v2.6-flash and pro versions both show outstanding technical potential. In the AI video and photo generation list, although Google and OpenAI occupy some of the top positions, many major generation models of domestic teams such as ByteDance, Alibaba, and MiniMax also performed steadily, showing strong global competitiveness as a whole. With large models