China

On August 21, 2026, DeepSeek’s API changelog launched experimental DeepSeek-V4-Flash-Vision-Exp (`deepseek-v4-flash-vision-exp`): JPEG/PNG/GIF/WebP via base64, URL, or Files API file_id, on Chat Completions, Anthropic-compatible Messages, and Responses. Text-agent scores match official V4-Flash; DeepSeek says multimodal-agent results jump toward Claude Opus 4.8. Distinct from the August 13 V4-Pro GA and price hike.

On August 14, 2026, Alibaba’s Qwen team posted Qwen3.8-27B on Hugging Face—a 27-billion-parameter native vision-language dense model with 262K context (extensible to 1M), thinking mode by default, and Apache 2.0 weights—the local companion to July’s 2.4T Qwen3.8-Max preview.

On August 13, 2026, DeepSeek said it will raise API prices for V4-Flash and V4-Pro and introduce peak and off-peak rates—increases Reuters put in a 50% to 1,100% range depending on model, token type, and hour—effective 16:00 UTC on August 16.

On July 19, 2026, Bloomberg reported Moonshot AI distributed a shareholder resolution seeking approval for a Hong Kong listing as early as within six months—capitalizing on Kimi K3’s frontier splash days after WAIC opened and global tech stocks reacted to Chinese open-path models.

On July 19, 2026, Alibaba’s Qwen team launched Qwen3.8-Max-Preview—a 2.4-trillion-parameter multimodal flagship live on Token Plan, Qoder, and QoderWork—claiming frontier-class capability second only to Claude Fable 5, with open weights promised soon but no public benchmark table yet.

On July 17, 2026, Chinese President Xi Jinping attended the opening of the World AI Conference and High-Level Meeting on Global AI Governance in Shanghai—his first in-person WAIC appearance—calling AI a ‘symphony of global cooperation,’ opposing overstretched national-security framing, and backing a new World AI Cooperation Organization with dozens of partner countries.

On July 16, 2026—the same day Moonshot AI launched Kimi K3—Arena.ai unblinded the model at #1 on the Frontend Code Arena with 1679 Elo, ahead of Claude Fable 5 and GPT-5.6 Sol in blind human preference tests, a 17-place jump from Kimi K2.6 and first place in six of seven frontend domains.

On July 16, 2026, Beijing’s Moonshot AI launched Kimi K3—a 2.8-trillion-parameter open-class multimodal model with a 1M-token context window, API live the same day, and full weights planned for July 27—claiming frontier coding and agent scores just behind Claude Fable 5 and GPT-5.6 Sol at far lower token prices.

On July 15, 2026, China’s Interim Measures for the Administration of Artificial Intelligence Anthropomorphic Interaction Services took effect—regulating continuous emotional, human-like AI companions with youth protections, disclosure, and content duties, while exempting ordinary customer service, Q&A, work assistants, and education tools that lack sustained emotional interaction.

On June 10, 2026, OpenAI published research on PRC-linked influence operations that used ChatGPT to generate content aimed at shaping U.S. public and political debate over AI infrastructure, data centers, and related trade issues such as tariffs. The company said it detected the campaign because operatives relied on its models to produce social-media-ready material, adding a new dimension to how AI tools themselves become instruments in information warfare around AI policy.

On April 25, 2026, Pony.ai announced a new-generation Level 4 autonomous driving domain controller developed in collaboration with NVIDIA, built on the NVIDIA DRIVE Hyperion platform and powered by DRIVE AGX Thor SoCs with NVLink interconnect, delivering up to 4,000 FP4 TFLOPS of compute performance.

On April 12, 2026, reporting highlighted how China, the U.S., Russia, and others have accelerated their development of artificial-intelligence-backed autonomous weapons and defense systems, with capabilities that reduce the need for human intervention in combat decisions. This buildup, compared to the dawn of the nuclear age, includes China's September 2025 military parade showcasing AI drones and the U.S. responding with its own advancements.

On April 6, 2026, reports emerged that OpenAI, Anthropic, and Google have begun collaborating through the Frontier Model Forum to detect and counter adversarial distillation by Chinese labs, sharing information on attempts to extract capabilities from US frontier models in violation of terms of service.

A consortium of Chinese research institutions and companies has released what they claim is the world's largest open multimodal dataset focused on physical interaction and robotics. The dataset includes synchronized vision, language, force, and proprioception data from real robots.

Alibaba's Qwen team released four compact open-weight models (0.8B, 2B, 4B, and 9B parameters) in the Qwen 3.5 family on March 1, 2026, optimized for local inference on smartphones, laptops, and other resource-constrained hardware while supporting native multimodal inputs.

On February 23, 2026, Anthropic published “Detecting and preventing distillation attacks,” alleging three Chinese AI labs—DeepSeek, Moonshot, and MiniMax—used roughly 24,000 fraudulent accounts to generate more than 16 million Claude exchanges for illicit capability extraction. The company framed the campaigns as ToS and regional-access violations and detailed detection and defense investments against industrial-scale distillation.

On February 16, 2026, Alibaba’s Qwen team released Qwen3.5-397B-A17B, a natively multimodal open-weight MoE model (397B total parameters, 17B active) with strong performance in reasoning, coding, agent capabilities, and vision-language tasks. It features an innovative hybrid architecture for high inference efficiency and a 1M context window in the hosted Plus version.

On February 12, 2026, Shanghai-based MiniMax released M2.5 and M2.5 Lightning. The models achieve near state-of-the-art results on coding, agentic tool use, and productivity benchmarks at roughly one-tenth to one-twentieth the cost of leading models like Claude Opus 4.6, with Lightning variant offering 100 tokens per second throughput.

On February 12, 2026, OpenAI sent the House Select Committee on Strategic Competition between the United States and the Chinese Communist Party a memo titled “Updated Stakes for American-Led, Democratic AI,” alleging DeepSeek used sophisticated, multi-stage pipelines—including obfuscated third-party routers—to harvest model outputs for distillation. The same-day letter framed industrial-scale free-riding as a national-security and competitive threat ahead of expected Chinese model releases.

On February 11-12, 2026, Zhipu AI (Z.ai) released GLM-5, a ~745 billion parameter open-weight frontier model. Trained entirely on Huawei Ascend chips with no NVIDIA dependency, it achieves competitive results on coding and agentic benchmarks, tops several open-source leaderboards, and triggered a significant stock surge for the publicly listed company.