AI large model weekly list: Claude-fable-5 continues to top the list, domestic models are stable in the middle and upper reaches

IT House reported on July 9 that the Arena (arena.ai) platform announced the latest weekly list of large model battles. The statistics in this issue cover June 29 to July 5, 2026. The top spot this week is still firmly occupied by 🇺🇸 Anthropic’s Claude-fable-5. The score slightly increased by 1 point. American manufacturers have an absolute advantage at the top. Domestic models remain stable overall, with many models ranking among the top 30. IT Home attaches lists in major fields as follows: Comprehensive list The overall pattern of this comprehensive list has not changed much. The top five are all Claude series models of 🇺🇸 Anthropic. The ELO score difference between them is within 15 points, which is within the statistical error range, and is regarded as the first echelon. The top ten are all occupied by American manufacturers, and the ranking changes are basically 1~2 places. Only 🇺🇸 Gemini-3.5-flash rose 2 places to 11th, which is the model with the largest ranking increase outside the top ten; 🇺🇸 glm-5.1 fell 3 places to 22nd, and the ranking change is relatively obvious. Domestic models are generally in the middle and upper reaches, and the echelon is relatively stable: Alibaba qwen3.7-max-preview ranked the highest, ranking 15th, with an ELO of 1475 points, up 1 place from last week; Baidu ernie-5.1 ranked 27th, with an ELO of 1468 points, the same ranking; Xiaomi mimo-v2.5-pro ranked 29th, with an ELO of 1466 points, the ranking remains the same. Rank Model Vendor Score (±CI) Votes Price ($/M) Input/Output Context 1 🇺🇸 Claude-fable-5 Anthropic 1509 ±9 4350 $10 / $50 1M 2 🇺🇸 Claude-opus-4-6-thinking Anthrop