Top Tech Stories
1
Top Story
Alibaba previews 2.4-trillion-parameter Qwen3.8 ahead of open-weight release
Alibaba claims the model ranks second behind Fable 5.
2
Moonshot AI pauses new Kimi K3 subscriptions over GPU limits
Subscriptions will split into Kimi and Kimi Code tiers.
3
David Sacks says Chinese AI solved bugs US models blocked
Kimi K3 resolved 15 vulnerabilities Codex and Fable declined.
4
David Sacks accuses OpenAI and Anthropic of seeking regulatory capture
He claims they use informal warnings against Chinese open-weight models
5
Alibaba repost suggests China mandates open-weight Qwen models
Commentators say the repost implicitly confirms a strategic pivot.
6
SpaceX reportedly orders 13,000 Nvidia GB300 server racks from Foxconn
Each server rack costs an estimated $4 million.
7
Paul Graham claims trial lawyers oppose autonomous vehicles to protect profits
Elon Musk agreed, calling the legal opposition terrible.
8
Conviction founder Sarah Guo backs vertical AI over foundation models
Her strategy targets specialized fields like clinics and courtrooms.
9
SpaceX targets Starship Flight 13 launch for Friday
The fully stacked vehicle will launch from Starbase, Texas.
10
Andreas Kirsch jokes about US border entry after online AI debate
The joke followed OpenAI's roon praising US free speech.
11
AI2's Nathan Lambert says Chinese developers favor Claude for coding
Developers bypass restrictions using local tools like Moonshot's Kimi.
12
Replit founder Amjad Masad questions Nvidia stock drop after Kimi K3
Masad argues the hardware manufacturer's shares should have risen.
13
Levent Alpöge's AI proof of Jacobian conjecture is a prank
He claimed a model named Fable 5 found a counterexample.
14
Garry Tan advocates "file over app" philosophy over transient AI
Tan highlights Markdown as a resilient, universal data format.
15
Claude Code's Thariq Shihipar says team cut system prompt by 80%
He also uses Slack to coordinate multiple Claude Code agents.
16
Under Secretary of War disputes origins of DeepSeek restrictions
Emil Micha cites a 2026 Congressional law instead.
17
Grok 4.5 tops VulcanBench coding benchmark with 91.3% score
It outperformed Claude Fable 5 in accuracy and cost.
18
Cloud AI guardrails blocked Hugging Face cyberattack analysis
Engineers completed forensics using an unrestricted local model.
19
ChatGPT Work runs locally on desktop, cloud-based on mobile
Local desktop execution allows direct access to system files
20
Elon Musk predicts AI will make all goods and services free
He replied to Bojan Tunguz's post about free intelligence