Synovus Financial weighs merger options after drawing interest - Bloomberg
Investing.com -- Alibaba (NYSE:BABA) has released Qwen3-Coder-480B-A35B-Instruct, its most advanced open agentic code model to date, featuring 480 billion parameters with 35 billion active parameters.
The new model natively supports a 256K token context length and can scale up to 1 million tokens with extrapolation methods. According to Alibaba, the model achieves top-tier performance among open models across multiple agentic coding benchmarks, including SWE-bench-Verified.
Alongside the model, Alibaba is open-sourcing Qwen Code, a command-line tool for agentic coding. The tool, forked from Gemini Code, includes custom prompts and function call protocols designed to maximize Qwen3-Coder’s capabilities.
The company trained the model on 7.5 trillion tokens with a 70% code ratio, enabling strong coding abilities while maintaining general and mathematical capabilities. The model underwent post-training with reinforcement learning on real-world coding tasks, which significantly improved code execution success rates.
Developers can access Qwen3-Coder through multiple interfaces, including Qwen Code, Claude Code, and Cline. The model is available through Alibaba Cloud Model Studio with API access.
Alibaba stated it is working on additional model sizes to deliver strong performance while reducing deployment costs. The company is also exploring whether the Coding Agent can achieve self-improvement.
This article was generated with the support of AI and reviewed by an editor. For more information see our T&C.