Evaluation Knowledge Intelligence

6.1 In the CompuTEX 2026 keynote speech at the Taipei International Computer Show, NVIDIA CEO Huang Renxun officially released the Vera processor. This is NVIDIA's first CPU specially designed for the AI agent AgenticAI. It is also NVIDIA's next major bet in the data center CPU field after Grace.

11.8 Source:httpscpb.com

5.8 来源httpswww.aicpb.comzh

12.20 Based on multiple dimensions—including general model capabilities, domain-specific capabilities, vendor execution capabilities, solution richness, strategy and innovation, and market influence—Omdia conducted a comprehensive assessment of the top 12 leading Chinese model vendors, categorizing them by score into three quadrants: Leaders, Challengers, and Aspirants.

6.26 On June 23, 2026, French AI unicorn Mistral AI officially released OCR 4, a document intelligence model. This is not just another OCR upgrade—it's a paradigm shift that turns documents from "walls of text" into "semantic maps," where every paragraph, table, equation, and signature is localized, classified, and scored for confidence.

6.11 DarioAmodi, co-founder and CEO of Anthropic, issued a serious warning in a policy article that the large-scale job loss caused by artificial intelligence is not a short-term economic adjustment or business operation problem, but a structural inevitable result determined by the AI technology itself that reproduces human cognitive capabilities.

10.12 Source：httpscpb.com

7.9 On July 9, OpenAI officially released the GPT-Live series of voice models, including GPT-Live-1 and GPT-Live-1 mini, rolling out globally to ChatGPT users. GPT-Live-1 is available to Plus, Pro, and Go subscribers, while GPT-Live-1 mini is for Free users, with API access planned soon.

8.31 On August 30, at the 2026 Medical AI Innovation Forum in Shanghai, Ruijin Hospital and Huawei Cloud officially released the RuiPath 2.0 pathology AI model. The new model features comprehensive upgrades across diagnostic capabilities, interpretability, rare disease data training, performance, and lightweight deployment. It has already been deployed in over 90 hospitals across China, marking a significant step in the large-scale clinical adoption of pathology AI.

6.1 This model is officially positioned as the first domestic open source flagship to have the three core capabilities of "cutting-edge programming and agent capabilities million-level context native multimodal", directly benchmarking GPT-5.5ClaudeOpus4.7 and Gemini3.1 Pro and other overseas closed-source top models

4.1 Large Language Models (LLMs) are rapidly evolving from simple text-prediction systems into advanced reasoning engines capable of tackling complex challenges. Originally designed to predict the next word in a sentence, these models have now evolved to solve mathematical equations, write functional code, and make data-driven decisions.

9.29 Link：https://www.superclueai.com/

5.25 While Western frontier labs are still struggling with trillion-dollar valuations and profitability, DeepSeek has proven an alternative path through extreme cost-performance: leveraging open-source, low-cost pricing, and a multi-model matrix to rapidly capture the developer ecosystem, then using ecosystem scale to feed back into technological iteration.

7.13 SenseTime has officially released and fully open-sourced the SenseNova-Vision unified vision model. Unlike previous approaches that packaged multiple expert models for detection, segmentation, and depth prediction

6.29 How do U.S. tech companies control AI costs amid exponential token consumption growth in 2026?

6.11 The ability to "long autonomous work" is particularly critical. It allows AI to transition from "assistive tools" that require human confirmation every time to "autonomous agents" that work independently for a long time.

7.3 On July 3, Meituan officially released its next-generation foundation model LongCat-2.0, featuring an MoE architecture with 1.6 trillion total parameters and approximately 48 billion activated parameters per token.

7.31 ​On July 31, ByteDance officially released its next-generation video generation model Seedance 2.5, doubling the single-generation length from 15 to 30 seconds, supporting up to 50 multimodal reference inputs and native 4K quality. The model is rolling out to J Dream and Doubao Pro, with API access to Volcano Ark coming soon.

6.2 It retains and enhances Qwen 3.7's complete agent capabilities in the use of text coding tools and productivity workflows. At the same time, it fully upgrades vision-language fusion capabilities, supports image, video, screen, web pages and text input, and can be completed in the GUI graphical user interface CLI command line interface and tool environment.

5.29 MistralAI, the European leader in open source models, officially released its new product Vibe on May 28, and simultaneously announced its comprehensive entry into industrial-grade AI applications and self-built data center strategies.