- GPT-4.1 significantly improved coding capabilities, achieving 54.6% in SWE-bench Verified tests, outperforming GPT-4o by 21.4% and GPT-4.5 by 26.6%[12][13][15] - GPT-4.1 demonstrated enhanced instruction-following ability, scoring 38.3% in Scale's MultiChallenge benchmark, a 10.5% improvement over GPT-4o[12][13][17] - GPT-4.1 achieved new SOTA in long-context understanding, scoring 72.0% in Video-MME benchmark, surpassing GPT-4o by 6.7%[12][13][22] - GLM-4-32B-0414 utilized 15T high-quality data for pretraining and applied reinforcement learning techniques to improve instruction-following, engineering code, and function-calling capabilities[26][28][30] - GLM-Z1-32B-0414 enhanced mathematical and logical reasoning through stack-sorting feedback reinforcement learning, significantly improving complex task-solving abilities[31][33] - GLM-Z1-Rumination-32B-0414 focused on deep reasoning and open-ended problem-solving, leveraging extended reinforcement learning and search tools[34] - Seed-Thinking-v1.5 adopted MoE architecture with 200B parameters, achieving 86.7% on AIME 2024 and 55.0% on Codeforces benchmarks, showcasing strong STEM and coding reasoning capabilities[35][37][41] - Seed-Thinking-v1.5 employed dual-track reward mechanisms for training, combining verifiable and non-verifiable data strategies to optimize model outputs[36][38][40] - GPT-o3/o4-mini introduced visual reasoning into the chain of thought (CoT), achieving 96.3% accuracy in V* benchmark, marking a major breakthrough in multimodal reasoning[42][46][48] - Video-R1 model applied T-GRPO algorithm to incorporate temporal reasoning in video tasks, achieving 35.8% accuracy in VSI-Bench, surpassing GPT-4o[63][65][68] - Pangu Ultra, a dense model with 135B parameters, achieved top performance in most English and all Chinese benchmarks, rivaling larger MoE models like DeepSeek-R1[69][73][74]

Market Overview - The AI and robotics sector experienced a sideways adjustment, with the Huaxia Sci-Tech AI ETF (589010) declining by 0.9% as of April 18. Key stocks such as Chipone Technology and Anlu Technology led the decline [1] - The Robotics ETF (562500) remained flat, with a trading volume of 415 million yuan and a turnover rate of 3.57%, indicating active market trading [1] Key Events - On April 19, the world's first humanoid robot half marathon was successfully held in Beijing, with 20 robot teams participating. The TianGong Ultra robot from the Beijing Humanoid Robot Innovation Center finished first, completing the 21-kilometer race in 2 hours and 40 minutes [1] - The CTO of the Beijing Humanoid Robot Innovation Center, Tang Jian, stated that the price of humanoid robots is expected to be comparable to that of an entry-level sedan as they become more widely adopted [1] Institutional Insights - Dongwu Securities highlighted several issues with humanoid robot hardware revealed during the marathon: 1. Insufficient battery life, as most robots required battery changes mid-race except for the walking robot [3] 2. Inadequate joint cooling, with robots needing cooling sprays to maintain joint functionality during the competition [3] - The event, hosted by local government and broadcasted by CCTV, garnered significant public attention, reflecting the government's strong emphasis on the humanoid robot industry. This focus is expected to accelerate the development of the humanoid robot sector and help overcome key challenges for commercialization [3] Popular ETFs - The Robotics ETF (562500) is the largest robotics-themed ETF in the market, facilitating investors' access to the Chinese robotics industry [3] - The Huaxia Sci-Tech AI ETF (589010) serves as the brain of robotics, with a 20% fluctuation range and small-cap elasticity, aiming to capture pivotal moments in the AI industry [3] Technological Developments - On April 18, Microsoft launched the BitNet b1.58 2B4T, the largest 1-bit AI model globally, featuring 2 billion parameters and capable of efficient operation on standard CPUs like Apple's M2. The model utilizes minimal weights of -1, 0, and 1 to achieve high memory and computational efficiency, outperforming similar models from Meta and Google in various inference tasks [2]