China’s AI labs in depth — DeepSeek, Qwen, Kimi, GLM and Doubao — covering model capability, open weights, pricing and how they compare with Western rivals.
Kuaishou’s AI video spinoff has released Kling 4.0, doubling maximum output to 30-second clips and adding granular keyframe control as it prepares for a standalone Hong Kong listing.
Anthropic has named Alibaba, Moonshot AI and DeepSeek in a disclosure about distillation campaigns aimed at its models. Here is what the reports say, what they do not say, and why the pricing gap explains the stakes.
DeepSeek AI has released the DeepSeek Harness into developer preview, an MIT-licensed agent framework where every component is implemented as a plugin. The release expands DeepSeek’s open-source developer tooling beyond its language models.
DeepSeek has warned of a ‘significant’ price rise, The Next Web reports, reversing the cheap-AI pitch that made it a favourite among cost-conscious developers.
Bloomberg reports Alibaba has released an AI model with performance rivalling Anthropic’s Claude. Here is what the launch means for developers, pricing and the intensifying AI race.
Axios reports that DeepSeek’s new low-cost model is accelerating AI’s race to zero on pricing. Here is why collapsing token prices matter for developers, rival labs and the economics of the entire model market.
DeepSeek has unveiled a public beta API for its flagship AI model, according to Bloomberg. The move gives developers direct programmatic access at a moment when the company is also reported to be building a large data centre in Inner Mongolia.
DeepSeek has reportedly told investors it is pausing a funding process following a wave of viral social media posts, according to reporting from Bloomberg and Fortune. Here is what is confirmed, what is not, and why it matters for teams building on DeepSeek models.
Nvidia chief executive Jensen Huang has defended Chinese AI development amid industry anxiety over Moonshot AI’s Kimi models, according to an Axios exclusive. Here is what the intervention signals for developers, investors and the open-weights ecosystem.
Tom’s Hardware reports that Z.ai, the developer of the GLM model family, has powered up a 1-gigawatt AI data center built entirely on Chinese chips, running multiple 10,000-chip clusters with zero Nvidia silicon.
Alibaba has previewed Qwen3.8 and, according to Quartz, claims the model rivals Anthropic’s Fable 5. With no benchmarks, pricing or release date yet disclosed, we examine what the claim would actually take to substantiate.
Forbes has published a comparison asking whether users should try Kimi K3, China’s latest AI model, against ChatGPT and Claude. Here is what the reports say, what remains unconfirmed, and how to evaluate the model on your own workloads.
