On March 6th, AIbase discovered that Tongyi, an AI application under Alibaba's AI To C business, has integrated Alibaba's latest open-source inference model, Qwen-QwQ-32B. It's now available on both mobile and web platforms.
Qwen-QwQ-32B is Alibaba's newly released powerful inference model, characterized by its balance of performance and efficiency. It achieves a significant leap in mathematics, code, and general capabilities, rivaling DeepSeek-R1 in overall performance. While maintaining strong performance, Qwen-QwQ-32B significantly reduces deployment costs, allowing for local deployment even on consumer-grade graphics cards.
Users can experience this through the "Qwen-QwQ-32B" agent interface on both the Tongyi app and web version.
Previously, the Tongyi app also launched the Qwen-QwQ-preview inference model via agents, allowing domestic users to experience the powerful capabilities of Tongyi Qianwen first-hand. Some netizens on social media pointed out its excellent code generation capabilities, stating that "macro generation works perfectly on the first try."
It is understood that in the future, the Tongyi app and web version will continue to integrate the latest models from Tongyi Qianwen, applying deep inference capabilities to its agent ecosystem. This will revolutionize the personal AI assistant experience, further improving user efficiency in life, work, and study.
Furthermore, AIbase also found that the Tongyi app's product interaction is subtly changing. The homepage now features a cute, friendly girl as the primary interaction object, and several practical functions, such as photo-based question answering, have been added. The existing agent ecosystem has also been further enriched.
