Alibaba Expands AI Portfolio With Qwen and HappyHorse Models

Alibaba has solidified its presence in the global artificial intelligence market with the release of its Qwen 3.7 Max model and the debut of a top-ranked video-generation tool. The Qwen 3.7 Max model has gained recognition for its performance in coding and agentic workflows, while the company’s new text-to-video model, HappyHorse 1.0, recently secured the top position on the Artificial Analysis leaderboard [1, 2].

Key takeaways

• Qwen 3.7 Max outperforms competitors like DeepSeek and Opus 4.6 Max in technical evaluations such as Terminal Bench 2.0 and MCP Atlas [1].
• Alibaba’s HappyHorse 1.0 model reached the number one spot on the Artificial Analysis text-to-video leaderboard shortly after its release [2].
• Qwen 3.7 Max offers significant cost advantages, with input costs at half and output costs at one-quarter of the rates charged by GPT 5.5 [1].
• The Qwen 3.7 Max model features a tenfold speed improvement in GPU kernel optimization, though it faces challenges with token verbosity during long tasks [1].

Technical Performance and Market Positioning

The Qwen 3.7 Max model has distinguished itself through high precision in complex, long-horizon tasks and agentic workflows [1]. In coding-specific benchmarks, the model has surpassed both Opus 4.6 and 4.7, though it remains behind OpenAI’s GPT 5.5 in general intelligence and writing quality [1]. Beyond its technical output, the model is designed for integration, supporting the Anthropic API protocol to assist developers in incorporating it into existing projects [1].

While the model is highly efficient in GPU kernel optimization, users have noted that its tendency toward verbosity in token generation can increase costs during extended operations [1]. This necessitates careful resource management for developers utilizing the model for large-scale, automation-heavy projects [1].

The Rise of HappyHorse and Competitive Pricing

Alibaba’s entry into the video-generation space has been marked by the success of HappyHorse 1.0. Although the model topped the Artificial Analysis leaderboard immediately following its debut, its connection to Alibaba was not publicly disclosed by the company until several days after its release [2]. The model is currently undergoing internal beta testing, with plans for a future API launch [2].

These developments occur alongside an aggressive pricing strategy from Chinese AI labs. By offering API pricing significantly lower than Western counterparts, firms like Alibaba are intensifying competition within the industry [1]. This shift in cost structure, combined with the emergence of high-ranking models from Chinese developers, is forcing established global players to re-evaluate their market positions [1, 2].

Why it matters