Core Insights - Ant Group has launched its first trillion-parameter model, Ring-1T-preview, on Hugging Face, showcasing advanced natural language reasoning capabilities [3] - The model achieved a score of 92.6 in the AIME 25 test, surpassing all known open-source models and Gemini 2.5 Pro, and is close to GPT-5's score of 94.6 [2] - In the CodeForces test, Ring-1T-preview scored 94.69, exceeding GPT-5 [2] - The model ranks first among open-source models in authoritative lists such as LiveCodeBench and ARC-AGI-v1 [2] Technical Details - Ring-1T-preview utilizes the Ling-2.0 MoE architecture and has been pre-trained on 20 trillion tokens [2] - The model incorporates the previously released icepop method and has undergone RLVR training within Ant Group's self-developed efficient reinforcement learning system, ASystem [2] - The AReaL framework of ASystem has been open-sourced, indicating a commitment to transparency and collaboration in AI development [2]

Core Insights - Ant Group has launched its first trillion-parameter model, Ring-1T-preview, entering the trillion-parameter model arena [1] - The model is designed for natural language reasoning [1] Company Summary - Ant Group has developed and open-sourced a significant AI model, indicating its commitment to advancing artificial intelligence technology [1] - The introduction of the Ring-1T-preview model positions Ant Group as a competitive player in the AI landscape, particularly in natural language processing [1] Industry Summary - The launch of trillion-parameter models represents a significant trend in the AI industry, showcasing advancements in model complexity and capabilities [1] - Natural language reasoning models are becoming increasingly important in various applications, reflecting the growing demand for sophisticated AI solutions [1]

Core Insights - Ant Group has open-sourced its first trillion-parameter model, Ring-1T-preview, which is the world's first open-source trillion-parameter reasoning model [1][2] - The model demonstrates exceptional natural language reasoning capabilities, scoring 92.6 on the AIME 25 test, surpassing all known open-source models and Gemini 2.5 Pro, and is close to GPT-5's score of 94.6 [1] - In the CodeForces test, Ring-1T-preview achieved a score of 94.69, exceeding GPT-5, and ranked first among open-source models in LiveCodeBench and ARC-AGI-v1 [1] - The model also showcased reasoning abilities in the IMO25 test, successfully answering the third question and partially answering others [1] Development Efforts - The Ant Ling team is actively working on post-training for the Ling2.0 family 1T language base to maximize the natural language reasoning potential of the trillion-scale base model [2] - The official version of Ring-1T is currently under training [2]