Qwen3.8 Max – Cost per task higher than Astra on Artificial Analysis (artificialanalysis.ai)

Alibaba has announced the release of Qwen3.8 Max (0902), a proprietary AI model that touts strong performance in intelligence evaluations while maintaining a moderate pricing structure. Released in September 2026, Qwen3.8 Max achieved a notable score of 45 on the Artificial Analysis Intelligence Index, which outpaces the median of 25 among comparable models. The model’s capabilities extend to multimodal input, accommodating both text and images, and it features an impressive context window of 984k tokens. However, it is characterized by slower processing speeds, generating output at approximately 40 tokens per second. This model's significance lies in its competitive pricing and its performance benchmarks, which may influence how developers allocate resources for AI applications. Priced at $2.00 per million input tokens and $6.00 per million output tokens, Qwen3.8 Max positions itself as a cost-effective option for tasks requiring extensive reasoning and analysis. Despite its verbosity with a total generation of 190 million tokens during evaluations, its slower operational speed could present challenges in time-sensitive applications. Overall, Qwen3.8 Max represents a blend of capability and affordability, potentially reshaping usage patterns within the AI/ML landscape.