Alibaba's Qwen team released Qwen-MM-Plugins, an open-source repository that packages multimodal capabilities as installable skills for six agent harnesses: Claude Code, Codex, Qoder, OpenClaw, Qwen C…
New checkpoints for the Qwen 3.8B family have been detected on a model hub, suggesting additional instruction-tuned or specialized versions, though the Qwen team has not officially announced them. The…
Moonshot AI released Kimi K3 in July 2026 as a 2.8T-parameter open-weight MoE model with a 1M token context window and native vision support, scoring 88.3 on Terminal-Bench 2.1, close to top closed mo…
In 2026, edge inference has advanced to the point where an 80B Qwen model runs in just 4.3GB of RAM on a MacBook Pro, and a 35B model runs on an iPhone 18 Pro. This is achieved through a combination o…
A new arXiv study (2608.00077v1) audits visual-token pruning in OCR-critical multimodal large language models (MLLMs), finding that accuracy alone misses failures where correct answers lack local toke…
The US lead over China in AI has essentially disappeared, according to an analysis of model releases and deployment trends. Chinese models like DeepSeek's R1 and V3 and Qwen now match or beat US open-…
Alibaba's Qwen team released Qwen3.8-Max, a 2.4-trillion-parameter mixture-of-experts model with 95 billion active parameters and a million-token context window, and demonstrated an autonomous 16-day …
Alibaba Group Holding Ltd.'s AI model, Qwen, autonomously coded for 16 consecutive days, with every commit publicly available on GitHub, according to a report by The New Stack. The AI, developed by Al…
DingTalk and Feishu have evolved into central hubs for enterprise AI, with DingTalk's AI Assistant Platform and Feishu's Intelligent Partner enabling conversational BI deployment directly within their…
Qwen 3.8 Max, a new API model, bills $2 per million input tokens and $6 per million output, but its cheapest reliable configuration is not the default. Testing through the Synthorai gateway revealed t…
Steve Berry warns that AI companies, like Uber before them, will raise prices once users are dependent, and credits China's Alibaba with releasing free open-source models like Qwen as a market check. …
Chinese-developed open-weight AI models accounted for 41% of all downloads on Hugging Face between February 2025 and February 2026, surpassing US-developed models at 36.5%, according to Hugging Face C…
Headroom, a new browser extension, provides a real-time visual gauge of context window usage in AI chat conversations, warning users before the AI forgets earlier information. It supports DeepSeek, Ch…
Alibaba's Qwen team unveiled Qwen3.8-Max, a 2.4-trillion-parameter language model with 95 billion active parameters, designed to autonomously complete complex tasks over extended periods. In testing, …
Alibaba's Qwen team released Qwen3.8-Max, a 2.4-trillion-parameter mixture-of-experts model, now generally available via API with open weights promised next week. The model scores 86.6 on Terminal-Ben…
Alibaba announced Qwen3.8-Max, pledging to open-source the weights of a Max-class model for the first time next week, alongside the release of Qwen3.8-27B. OpenAI's super PAC appears to be running an …
A study from arXiv:2607.28814v1 found that preference-optimizing large language models (LLMs) for motivational interviewing by penalizing confrontation reliably lowers goal persistence across all test…
Anthropic, an AI company, accused Alibaba's Qwen AI lab of orchestrating the largest known model distillation campaign, using roughly 25,000 fake accounts to generate over 28.8 million exchanges with …
Alibaba's Qwen team unveiled Qwen3.8-Max-Preview, a 2.4 trillion-parameter multimodal model, at the World Artificial Intelligence Conference in Shanghai on July 19, claiming it trails only Anthropic's…
Alibaba's Qwen team released Qwen3.8-Max, a 2.4 trillion parameter multimodal AI model that trails only Anthropic's Claude Fable 5 in internal benchmarks, with open weights scheduled for release next …