华为发布麒麟 9050 系列实现首个 30B MoE 入端
10/13/2026 — 10/15, 12:10·3 sources (all)·4 reports (all)
Key takeaways
- Huawei unveiled the Kirin 9050 series at its 2026 media briefing, positioning it as the first native AI phone platform. Powered by a new DaV…
Huawei unveiled the Kirin 9050 series at its 2026 media briefing, positioning it as the first native AI phone platform. Powered by a new DaVinci architecture NPU, it achieves the industry's first on-device 30B MoE model with 2B active parameters.
Latest turnSecurity research team XBOW announced that they used GPT-5.6 Sol and other models alongside four researchers to build a full-chain Chrome exploit and breach the sandbox in just six days. Google has fixed the vulnerabilities and awarded a $250,000 bounty, demonstrating AI's practical capability in complex security vulnerability discovery.
Related AI tools
- Heat index
- 141
- All sources
- 3
- All reports
- 4
- First seen
- 2 days ago
Reports on this story headlines open the original
Security research team XBOW announced that they used GPT-5.6 Sol and other models alongside four researchers to build a full-chain Chrome exploit and breach the sandbox in just six days. Google has fixed the vulnerabilities and awarded a $250,000 bounty, demonstrating AI's practical capability in complex security vulnerability discovery.
IT之家AI score 76At the FrontKon conference, a developer explored running AI models directly in the browser, highlighting how WebGPU can accelerate parallel computations by tens of times. The talk demonstrated using WebAssembly and WebGPU for advanced in-browser workloads, though adoption remains low.
DEV Community · AIAI score 57
llama.cpp's WebGPU backend now supports streaming MoE experts from disk directly in a browser tab. A live recording demonstrates the setup running smoothly on a MacBook Pro M4 Pro.
Reddit · LocalLLaMAAI score 61Huawei announced the Kirin 9050 series at its 2026 media briefing, positioning it as the first native AI Phone platform. Powered by a next-gen DaVinci architecture NPU, the series achieves the industry's first on-device 30B MoE model with 30B total parameters, 2B active parameters, and under 2.5GB memory footprint.
IT之家AI score 65
Other stories people are talking about
- Heat index 735Google rolls out Gemini 4 Argon globally18 sources
- Heat index 360Meta launches Muse Wonder and Muse Wonder Full models11 sources
- Heat index 346Anthropic Study Warns Small AI Models Could Spread Across Devices10 sources
- Heat index 346Google Play Launches Personal AI App Execution Feature9 sources
- Heat index 333Character.AI Files for Chapter 11 Bankruptcy and Plans Asset Sale to Disney10 sources
How is heat calculated?About the methodHide
Heat counts how many independent sources covered a story in the last 48 hours: one source counts once no matter how many posts it published, decaying with a 24-hour half-life. What ranks first is what many people are talking about.
This page aggregates public feeds. Headlines and summaries are machine-organized and remain the property of the original authors; verify important facts at the source. If you believe a headline or summary infringes your rights, email the contact address in the footer with the page URL and basis of your claim, and we will remove or replace it after review.
- Surge
- Discussion rising fast
- New
- First report within 6 hours
- Rising
- Still gathering discussion