Tracked topic
Qwen3 is an AI model family developed by Alibaba, released as a set of large language models for natural-language tasks.
… In fact, Alibaba's Qwen-Audio-3.0-Realtime Plus now leads the Artificial Analysis Speech-to-Speech Index with a score of 84.1%, ahead of OpenAI’s GPT-Realtime-2.1 High at 79.1%. It topped the benchmark’s speech reasoning, conversational dynamics, and agentic performance tests. …
… DeepSeek helped kick-start this wave, but models such as Moonshot AI's Kimi K3 , GLM 5.2, and Alibaba's Qwen 3.8 Max are now delivering performance close to frontier US models at a fraction of the cost. …
… Qwen3.8-Max is a massive mixture-of-experts model with 2.4 trillion total parameters. However, only 95 billion parameters are activated per request to deliver strong performance while reducing inference costs and response times. …
… In addition to that, the team relocated skill management to the settings UI, improving how the application communicates errors regarding those skills, and updated the OpenCode model roster to support DeepSeek V4 Flash, MiniMax M3, Qwen 3.7 Plus, and Nemotron 3 Ultra Free. …
… With ACP, the agent Claude Code, Gemini CLI, Qwen Code, or whatever handles the "thinking" while the IDE provides the agent with context your open files, the codebase , displays proposed code diffs visually to the user, renders markdown, and prompts you for permission before the agent executes a te… …
… This was on Qwen 3.6 35b. …
… The company will need local vendors, and reports said that Alibaba's Qwen models will power text and image features. …
… When tested, the MacBook Air outperformed in most tasks, such as file extraction, Word performance, and Premiere Pro. …
… Massive performance drop: Ubuntu engineers warned that an upcoming 7.0.0-28.28 kernel release can lead to a substantial 43 times performance hit on certain compute-heavy workloads. …