A cloud architect benchmarked DeepSeek, Qwen, Kimi, and GLM over six weeks using Global API's unified endpoint, finding DeepSeek V4 Flash the price-to-performance champion at $0.25 per million output …
A new guide from an unnamed author details how to run Qwen locally on Windows and access it remotely from any device using Ollama, LiteLLM Proxy, and Tailscale. The setup was tested with a Qwen 3.5 9B…
Chinese open-weight AI models now cost 60 to 90 percent less per token than leading US closed models, with DeepSeek charging $0.435 per million input tokens and $0.87 per million output tokens versus …
Lawrence Architectures released UL-SMF, an open-source KV-cache compression fabric that reduces memory footprint by up to 384x with over 94% semantic retention, addressing the memory bottleneck in lon…
A developer spent $47 testing nine multimodal AI APIs from Chinese labs, including Qwen, GLM, Hunyuan, and Doubao, to find cost-effective alternatives to OpenAI's GPT-4o for image-heavy workloads. The…
A developer who recently graduated from a coding bootcamp shares cost-saving techniques for AI APIs, including model routing and response caching, which cut their API bill dramatically. The developer …
Alibaba released Qwen3.8-27B under the Apache 2.0 license and opened the weights for its flagship Qwen3.8-Max model on August 17, according to CNBC. The 27B model has 27,781,427,952 BF16 parameters an…
Alibaba launched a laptop-ready AI model and released the weights of its most powerful Qwen model, escalating its rivalry with Meta in open-weight AI. The new model is designed to run on laptops, and …
Hugging Face data shows only one model appeared on both the top-25 most-downloaded and most-liked lists in 2026, with All-MiniLM-L6-v2, a 2021 Sentence Transformers model, downloaded 1.55 billion time…
On July 15, 2026, the Cyberspace Administration of China registered Apple's generative AI service, making Apple the first foreign company authorized to offer a proprietary AI model in China. According…
Nvidia has pre-booked TSMC's 1.6-nanometer A16 process capacity for its second-half 2028 Feynman GPUs, locking up advanced packaging and silicon photonics to deny rivals the manufacturing capacity nee…
The U.S. State Department is preparing a letter telling the 35 signatories of its AI Opportunity Statement, including Japan, Australia, and South Korea, that they cannot join both Washington's Pax Sil…
Qwen 3.8 27B, an open-source model from Alibaba, exhibits a tendency to overthink simple queries, generating excessive chain-of-thought reasoning that inflates token usage and latency. The behavior st…
Apple has reportedly trained a large language model specifically for the Chinese market with support from Alibaba, according to people familiar with the effort. The model would complement Apple's earl…
A Reddit user proposes specializing small language models (sub-16B) for dedicated tasks like math, medicine, and writing, orchestrated by a general Mixture-of-Experts model, to reduce cost and complex…
Hermes Agent, an open-source agent framework, can be paired with a locally hosted Qwen model via Ollama or LM Studio to create a zero-token-cost AI agent stack, eliminating per-token API fees. Key con…
On August 14, Alibaba's AI arm Qwen released Qwen3.8-27B, a compact model that is byte-for-byte the most powerful ever released and small enough to run on a well-equipped PC or Mac, marking the arriva…
Alibaba's Qwen research lab released Qwen 3.8 27B, an Apache 2 licensed 27B parameter vision-capable LLM, which defaults to an 'xhigh' reasoning effort that causes it to overthink simple tasks, taking…
A developer tested DeepSeek, Qwen, Kimi, and GLM through Global API's unified endpoint, comparing them on price, performance, and features. The developer found DeepSeek V4 Flash the best overall budge…
Alibaba's Qwen team released Qwen3, a new large language model available in sizes including a 27B version, with official BF16 and FP8 weights on Hugging Face and community GGUF and MLX quantizations f…