Apple Intelligence has been approved by China's Cyberspace Administration, allowing the AI system to launch on iPhones in the country for the first time. The on-device generative AI service will use m…
ByteDance told Doubao users on 15 July that custom AI agents built on the platform had stopped working, leaving a read-only archive accessible until 15 October, after which data will be handled under …
Alibaba has integrated its Qwen AI model, known as Tongyi Qianwen, into Apple Intelligence for Chinese users, enabling on-device text comprehension, image understanding, and content generation across …
ByteDance has disabled Doubao's user-built agent feature, affecting 350 million monthly active users and over 8 million custom agents, after China's new Interim Measures for anthropomorphic AI interac…
A freelance developer tested four Chinese AI models—DeepSeek, Qwen, Kimi, and GLM—for a client dashboard requiring text summarization, aiming to reduce API costs. DeepSeek V4 Flash emerged as the top …
PrismML has unveiled a compressed version of Alibaba's Qwen model that slashes memory requirements by up to 15 times, a technical leap that could accelerate Apple's AI ambitions by enabling more sophi…
Alibaba's Qwen AI model will be integrated into Apple Intelligence, bringing AI-powered text and image understanding and content generation to users in China across iOS, iPadOS, macOS, and visionOS. T…
A developer built an n8n agent that reads a lead's website before writing cold emails and refuses to write when it cannot find sufficient evidence. The workflow fetches the lead's homepage, extracts b…
A proof-of-mechanism study of ontology-amplified distillation found that a Qwen3.6-27B student model, fine-tuned on 47 synthetic English-language preference pairs using Foundation AgenticOS ontology, …
ByteDance's Doubao, China's second most-used AI app with 350 million monthly active users, switched off its user-built agent feature at midnight Beijing time today, affecting over 8 million custom age…
Apple is in exploratory talks to acquire PrismML, a Caltech spinout that uses ultra-low-bit compression to shrink large AI models by up to 14x, enabling them to run natively on iPhones. PrismML raised…
A study analyzing ChatGPT usage patterns found that 71% of user queries could be handled by a capable local model, yet most organizations still route every request to a cloud API, incurring unnecessar…
Apple Inc. is in early talks with PrismML, a Khosla Ventures-backed startup spun out of Caltech, to use its compression technology that shrinks AI models from 54 GB to under 4 GB, enabling all 27 bill…
A developer built QuotePilot, a Qwen-powered app that automates cross-border B2B price quotes, using Qwen models to write most of the code for under $1 in API costs. The pipeline uses three Qwen model…
A developer cut their AI API bill by 95% by matching models to tasks, routing requests through cheap tiers first, caching aggressively, compressing prompts, and batching. The developer built a tiered …
AI is now involved in every stage of a cyberattack, from identifying targets to stealing data, according to research from Check Point. Criminal groups have been breaching government agencies using AI …
Bonsai 27B, based on Qwen3.6 27B, is the first 27B-class model to run on a phone, with a 1-bit variant at 3.9 GB fitting within an iPhone 17 Pro's memory budget. The model retains 90% of full-precisio…
A new measurement study by an independent developer using the open-source HomeLab Monitor tool found that running local large language models on a single NVIDIA RTX 3090 can cost as little as €0.046 p…
An open-source pipeline called ai-trains-ai uses nested reinforcement learning loops where an outer agent is RL-trained to write inner training jobs for small models, with the entire system run for ro…
A data scientist evaluated 10 AI coding models across 5 tasks, finding that DeepSeek-R1 scored highest at 9.4 but cost $2.50/M tokens, while Qwen3-Coder-30B offered the best balance at 8.8 and $0.35/M…