The AI video generation market in 2026 is defined by intense competition. ByteDance’s Seedance 2.5, MiniMax H3, PixVerse, Kling, and others are all pushing strong models. The interesting story is not which is best on benchmarks — it is how the open-source vs proprietary split is shaking out. For users, the practical implication is more … Read more
DeepSeek recently released the official version of V4 Flash, an open-source AI model optimized for Agent tasks. The interesting twist: the model architecture and parameter count are unchanged from the Preview version. What changed is post-training, and the result is meaningful gains in Agent-related benchmarks. The release matters for anyone building with open-weights models, particularly … Read more
On August 3, 2026, Alibaba released Qwen3.8, a new foundation model with 2.4 trillion total parameters. It placed second on the Arena leaderboard, behind only Anthropic’s Claude series, with particular gains in coding and professional agent tasks. The API is live on the Qwen platform, and the team has integrated it with the Qwen Office … Read more
Tencent’s Hunyuan team has released Hyra, a research agent built on their Hunyuan-3 model. The headline: Hyra solved a 50-year-old open problem in additive combinatorics — a question first posed in 1969 about the expansion behavior of a specific integer set. The interesting thing about this release is not just the result — it is … Read more
On August 3, 2026, OpenAI co-founder Greg Brockman posted a short but pointed message about “real-time audio” and a new architecture direction. The post does not announce a product, but it signals something more interesting: OpenAI is publicly talking about the limits of the Transformer architecture and what comes next. The post is brief and … Read more
Zhipu has released GLM-5.2, the latest version of its flagship open-source model. The model is now on Hugging Face, and it placed fifth globally on the most recent usage rankings, behind four other Chinese models in the top five. GLM-5.2 is positioned for long-context tasks, with a 1M token context window. For developers and businesses … Read more
Baidu has upgraded its “Da Zai” (搭子) agent platform, adding an enterprise version and announcing a Skill access standard for connecting enterprise AI systems. The move is part of a broader pattern in 2026 of Chinese tech companies pushing into agent infrastructure. For developers and businesses looking at the Chinese agent ecosystem, Baidu’s Skill standard … Read more
Cloudflare has launched a Billable Usage API that gives developers structured access to per-product, per-time-period usage data. For teams running on Cloudflare’s stack — Workers, Workers AI, R2, and the rest — the API changes how you think about cost visibility. The shift is small in API terms but meaningful for how teams manage cloud … Read more
Saner.AI, a Chinese brain-computer interface company, has completed a 3.3 billion yuan angel round, the largest such round in the Chinese BCI space to date. The company is focused on invasive BCI products, with funding going to clinical trials and team expansion. The round is a signal that the BCI space in China is getting … Read more
According to recent OpenRouter usage data, Xiaomi’s MiMo model has moved into the global number two position for AI model usage, behind only DeepSeek-V4-Flash. The data, cited by 36Kr based on OpenRouter’s weekly figures, shows the top five models in usage are all Chinese. For the AI model market, this is a significant data point. … Read more