#mobileAI
Mobile LLM inference still faces substantial efficiency gaps across CPUs, GPUs, and NPUs, according to a new cross-layer benchmarking study using the PowerBench profiling tool.

#MobileAI #EdgeAI #LLM #Hardware
https://arxiv.org/abs/2607.05475
October 20, 2026 at 4:01 AM
SchemaWeaver introduces schema-guided parallel decoding to cut LLM tool-calling latency on mobile devices, producing multiple tokens per decode pass without retraining. It targets memory-bandwidth limits in the decode phase.

#LLM #MobileAI #OnDeviceAI
https://arxiv.org/abs/2610.17436
October 14, 2026 at 12:01 PM
China Mobile bundling tokens with Honor handsets is a genuine first — telco billing reaching edge inference, no app store gatekeeper. But no published latency or routing detail makes the bundle hard to price against a plain API call. Mid-range Honor users: I doubt it beats per-call in practice.
China Mobile and Honor launch token package to lower AI barriers
China Mobile and Honor have launched the MobileAI Expert Plan, a bundled token-and-connectivity package starting at 169 yuan a month, which pairs Honor's AI smartphones with China Mobile's AI services and phone-bill payment. Honor CEO Li Jian says the Magic 9 series is selling strongly, but warns that soaring memory and component costs are squeezing margins across the industry.
aichina.news
October 13, 2026 at 1:07 PM
Vals AI ranks Claude Opus 5.5 first on Terminal-Bench 4.0 at 65.15%.
You don't need a laptop to reach it. SmophyAI works in a mobile browser with the same models, so a prompt between meetings goes to Multi-Chat from your phone. Start free.
#AI #MobileAI #GenerativeAI
October 10, 2026 at 6:30 PM
GHARP animates 3D human heads in real time from a few images and a driving expression signal, splitting offline identity reconstruction from lightweight on-device animation. The animation stage runs a small network…

#GaussianSplatting #MobileAI #RealTimeRendering
https://arxiv.org/abs/2610.10945
October 10, 2026 at 12:01 AM
OpenPhone proposes a device-cloud collaboration system for mobile GUI agents, pairing an enhanced 3B on-device model with cloud models to balance cost and capability. The approach uses two-stage SFT then GRPO training on…

#OpenSourceAI #MobileAI #GUIAgents #MLLM
https://arxiv.org/abs/2510.22009
October 7, 2026 at 4:01 AM
October 7, 2026 at 11:08 AM
New work introduces a 2.7-bit quantization framework for running vision-language models on mobile hardware, compressing Llama 3.2 11B Vision Instruct to 3.7 GB for Arm CPU inference.

#Llama #VLM #MobileAI #Quantization
https://arxiv.org/abs/2608.21134
October 5, 2026 at 6:01 AM
The smartphone leaks are getting out of hand. Is Billy Smith’s secret AI interface about to completely render traditional app stores obsolete overnight? Here is what Silicon Valley is trying to hide from us. 📱💥 #TechNews #MobileAI
billy smith
Click to read full live report and analysis.
eslammsd-mobile-games-zone.static.hf.space
October 4, 2026 at 10:23 AM
King vs. Green just shattered the mobile tech space. Is this the end of traditional smartphone design as we know it? 🔥📱 #TechNews #MobileAI
кинг, грин
Click to read full live report and analysis.
eslammsd-mobile-games-zone.static.hf.space
October 4, 2026 at 1:25 AM
A new arXiv paper proposes a compute-storage co-design for mobile NPU LLM serving, reconciling KV cache reuse with statically compiled graphs under tight memory constraints. The work targets on-device inference where traditional…

#AI #LLM #MobileAI #Semiconductors
https://arxiv.org/abs/2609.34727
September 29, 2026 at 10:01 PM
New arXiv paper introduces GUI-Hopper, which augments mobile GUI agents with app-native deeplinks discovered through static analysis and validated on real devices. The hybrid approach replaces multi-step tap sequences with single…

#AI #MobileAI #GUIagents #HCI
https://arxiv.org/abs/2609.30887
September 28, 2026 at 4:01 PM
⚡📱 Speed News, Smartphones & GenAI – New in our #MobileAI special issue:

@SubinPaul on how rural "news informers" in India use ChatGPT on low-cost Android phones to condense stories into punchy lines.

The result: a deeper "ruralization" of Hindi news.

🔗 doi.org/10.1177/2050...
September 28, 2026 at 3:12 PM
When football strategy meets next-gen mobile AI, everything changes on the pitch. Once Caldas is pioneering real-time match analytics right from smartphones—and the results are wild. ⚽📱 #OnceCaldas #MobileAI #TechInSports
once caldas
Click to read full live report and analysis.
eslammsd-mobile-games-zone.static.hf.space
September 26, 2026 at 2:07 AM
September 23, 2026 at 8:49 AM
🌱📱 Greening Mobile AI? New in our #MobileAI special issue: Gerard Goggin asks what happens to the environment as mobile media and AI become entangled.

Decades of "green phone" effort, little progress – and AI phones, wearables & cloud add new burdens.

🔗 doi.org/10.1177/2050...
Greening Mobile AI? - Gerard Goggin, 2026
Mobile AI is very much a work in progress. No better time at the outset, then, to consider its environmental implications. As mobile media and communication are...
doi.org
September 22, 2026 at 8:03 AM
📈 Overall AI Market: CAGR 18%-35% (2026–2035)
🤖 Generative AI: 22.9%
🩺 Healthcare AI: 19.1%
🔒 Cybersecurity AI: 22.5%
📱 Mobile AI: 25.1%
#AIGrowth #GenerativeAI #HealthcareAI #CybersecurityAI #MobileAI
View in Timelines
September 21, 2026 at 2:00 PM
2-nanometer, AI-boosted performance: Dimensity 9600 Pro raises the bar across CPU, GPU, and memory. #Dimensity9600Pro #MediaTek #2nm #MobileAI #Performance #GPU #Tech thedailytechfeed.com/mediatek-deb...
September 15, 2026 at 3:48 PM
"Tra qualche anno non ripareremo più lo smartphone: anticiperemo l'usura." Sentita ieri al bar tech.

- Manutenzione da reattiva a predittiva
- Onde a 165Hz contro umidità e polvere
- App come Speaker Cleaner già tracciano la via

#TechTrends #MobileAI

https://smartlinks.live/l/y3V361
September 5, 2026 at 8:17 AM
🤖 AI Apps: ~39% CAGR
📈 AI Computing Power: 39.8%
🎥 AI Video Analytics: 31.8%
🛍️ AI in Retail: 24.3%
📱 Mobile AI: 22.5%
#AIGrowth #AIApps #AIComputing #AIVideo #AIRetail #MobileAI
View in Timelines
August 17, 2026 at 2:00 PM