AI Model Price War: DeepSeek V4 Flash 21x Cheaper

Three flagship AI models, Grok 4.6, Qwen3.8-Max, and DeepSeek V4 Flash 0731, launched this summer, showing a significant price difference. DeepSeek V4 Flash 0731 undercuts the other two models by more than 20 times on output token pricing. Grok 4.6, released on August 12, is a closed, pay-per-token model. Alibaba's Qwen3.8-Max, available since August 3, is an open-weight model with 2.4 trillion parameters. DeepSeek V4 Flash 0731, out on July 31, offers a much lower price point. What's interesting is that Grok 4.6 leads in composite intelligence. Qwen3.8-Max balances open weights with performance. DeepSeek V4 Flash 0731 wins on cost, especially for high-volume tasks. These models arrived within weeks of each other, highlighting the rapid pace of AI development. This speed means that for Canadian developers and engineering teams, choosing an AI model now involves more than just intelligence scores; it's also about license terms, context budgets, and pricing. The bottom line is that these factors can change a monthly API bill by thousands of dollars.