InAgent achieves 90.2% on OSWorld: the computer-use agent gap narrows for the Chinese stack
A Chinese CUA becomes the first to surpass 90% on OSWorld. The gap with US frontier labs is now a harness gap, not a model gap.
6 h ago 5 6
A Chinese CUA becomes the first to surpass 90% on OSWorld. The gap with US frontier labs is now a harness gap, not a model gap.
6 h ago 5 6
MiniMax releases H3, an open-source full-modal (text, image, audio, video) model at one-third the price of competitors. The AI video market stops being binary: expensive closed-source or weak open-source.
Jul 31, 2026 at 12:50 12 14
The Pragmatic Engineer publishes a "deepdive" on Anthropic's internal engineering. The surprising read: "ever more" code review and testing by AI, but the two-pizza teams hold. Smoothing, not flipping.
Jul 28, 2026 at 20:57 8 10
Black Forest Labs releases FLUX 3, its new generation of multimodal flow matching models. The lab claims superior results to Seedance 2.0 and Gemini on internal benchmarks. It remains to be confirmed by third-party evaluations.
Jul 24, 2026 at 21:00 8 8
Days after Kimi K3 went live, Moonshot has paused new paid subscriptions - a rare public admission that a Chinese frontier lab hit its own inference ceiling on launch week.
Jul 19, 2026 at 23:50 14 13
Moonshot has priced Kimi K3 at 3-15× the tokens of DeepSeek - the first time a Chinese frontier lab has publicly asked for a premium against its domestic peers, and a break with the price-war narrative.
Jul 18, 2026 at 11:11 20 9
Zhipu would generate $1 billion in annual revenue according to leaks. First Chinese frontier lab with measurable commercial traction at this scale.
Jul 17, 2026 at 09:19 29 8
The first open-weights model from Thinking Machines Lab (Mira Murati) is here: Inkling, MoE 975B params (41B active), multimodal, Apache-2.0 license. The most discreet box in the frontier scene enters the open market through the front door.
Jul 16, 2026 at 21:46 25 8
Anthropic is preparing a US listing as early as October, according to Tech in Asia. The number nobody has yet: at what price the "safety-first" narrative sells to public investors when OpenAI still sets the tempo.
Jul 16, 2026 at 08:46 14 8
Liang Wenfeng's open-weight lab discusses with advisors; targeted listing in 2027, valuation still unknown.
Jul 15, 2026 at 10:14 23 7
The X-Square Robot laboratory publishes a three-layer "foundation stack" for embedded AI. It's not yet a robot GPT - but the architecture reading, it, is ripe.
Jul 13, 2026 at 17:27 3 11