Qwen3.8-Max Release: Scores Slightly Higher Than DeepSeek-V4-Flash, Priced at Nearly 20x

BlockBeats News, August 3rd, Alibaba released Qwen3.8-Max official version. This is a 2.4T-parameter multimodal flagship model, focusing on Programming Agent, Professional Office, and Long Task, and the API has been launched.

In the four tests of TerminalBench 2.1, DeepSWE 1.1, NL2Repo, and Toolathlon Verified, it is only 1.7 to 3.9 points higher than DeepSeek-V4-Flash. The two tests have different frameworks and cannot be strictly compared head-to-head, but the difference is indeed small.

However, the price difference is very obvious. Qwen3.8-Max costs $2 for every million Token input and $6 for output; DeepSeek-V4-Flash only costs $0.14 and $0.28 respectively. In a comprehensive calculation, the expensive Qwen is about 20 times more costly.

Looking only at this performance and price, Qwen3.8-Max lacks cost-effectiveness. It uses nearly 20 times the cost to achieve only a few percentage points of lead.

The weights of Qwen3.8-Max will be open next week, and Qwen3.8-27B will also be open sourced at the same time.

Click the original link below to join the BlockBeats Beating · Feishu AI News Channel, monitoring global AI hotspots and news 24/7.