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.
5 h ago 5 6
Laguna 118B runs on a DGX Spark, Inkling goes multimodal, K3 secures its commercial license: benchmarks and clauses recap #23
Three architectures to dissect side-by-side: Poolside's MoE 118B / 8B active on a single machine; Inkling's multimodal 975B / 41B, also released as 276B / 12B; and Kimi K3's 2.8T / 104B context with 1M tokens, whose commercial agreement clause could exclude US companies.
17 h ago 10 10
"Cancelling Cursor": Quality debt takes precedence over feature velocity
Alex Yumashev (Jitbit) explains why he unsubscribed from Cursor: VSCode base 7-8 months behind, 7 bugs reported in a year and never fixed, Agent mode forced with every update, ghost support. An individual signal that perfectly illustrates the editor-orchestrator divide.
17 h ago 11 10
August 2, 2026: The AI Act GPAI becomes enforceable – day one of a still under-equipped regime
Since today, transparency in training, disclosure of protected sourcing, and systemic risk management are no longer recommendations but legal obligations for general-purpose AI models. The question remains whether the AI Office, barely established, will be able to translate this mandate into case law—under the explicit threat of an American response.
17 h ago 9 12
MCP goes stateless: the spec finally clears the last enterprise blocker
A new Model Context Protocol revision drops the stateful session requirement - the piece that made MCP awkward to scale behind a load balancer. Small change on paper; large one in deployment.
Jul 30, 2026 at 19:38 17 17
UK AISI and CAISI publish first joint cyber assessment of Kimi K3
The British (AISI) and American (CAISI) AI security institutes publish their first joint preliminary assessment of Kimi K3's cyber capabilities. Strong signal: open-weight frontier models fall within the scope of government evaluation.
Jul 26, 2026 at 18:40 10 12
Nvidia's $1.5B Amkor deal: the US packaging leg finally gets its capex
Nvidia signs $1.5B with Amkor for advanced packaging in the United States. The compute AI bottleneck rises one notch towards the sovereignty of the back-end.
Jul 26, 2026 at 18:40 8 9
Korea-US $950B: the AI supercycle signs its most political contract yet
Three days of meetings in Seoul and San Francisco result in a Korea-US package valued by ETNews at ~$950B. The `korea-ai-supercycle` thread moves from signal to lock.
Jul 26, 2026 at 18:40 14 11
UK AISI and CAISI publish first joint cyber assessment of Kimi K3
The NIST website hosts a preliminary joint assessment by UK AI Safety Institute and CAISI (US) of Kimi K3's cyber capabilities - the first formal cross-Atlantic evaluation of a Chinese open-weight frontier model.
Jul 25, 2026 at 10:27 10 11
"State-owned AI won't solve inequality": Rest of World's bold thesis on AI in the Global South
Rest of World investigates the national AIs launched by several countries in the Global South. The finding is stark: nothing competitive has emerged, and the sovereignty argument masks a repetition of the telecom errors of the 1990s.
Jul 24, 2026 at 21:02 8 12
"AI communism": Kimi K3 shakes the thesis of the model monopoly on Wall Street
On the TechCrunch Equity podcast, American VCs refer to the Kimi K3 effect on valuations as "AI communism." The real signal: the first time a Chinese open-weight model forces American CFOs to recalculate their P&L.
Jul 24, 2026 at 21:00 5 8
US House vote to restrict Chinese humanoid robots: the militarization of dual-use
The debate was shifting from industrial robotics to the geopolitical stage; the House vote officially places it there, with a cascading effect on Unitree, AgiBot, and XPeng.
Jul 24, 2026 at 09:33 11 8