[KongchangAI]
ModelQwen3.8 Max

Qwen 3.8 Max

An ultra-large-scale Mixture-of-Experts model released by Alibaba's Qwen team, with 2.4 trillion parameters and published with open weights. Its overall capabilities are comparable to leading closed-source models.

Timeline (last 90 days)

Oct 11

Alibaba revealed that Qwen 3.8 Max has undergone more than one month of automated self-improvement cycles, completing 33 iterations.

Unverified50%
Sep 30

The open-source model MiniMax M2.6 Pro scored 46 points, while GLM 5.3 and Qwen 3.8 Max scored approximately 45 points, both close to GPT-6 Sol's score of 48 points.

Unverified50%
Sep 30

On the AI Builder model intelligence leaderboard, Claude Opus 5.5 Max topped the rankings with 58 points, Xiaomi MiMo V2.6 Pro ranked 9th with 46 points, and Qwen 3.8 Max ranked 10th with 45 points.

Unverified50%
Sep 12

The reviewer believes that Qwen 3.8 Max performed comparably to Fable 5.1 on the same skateboarding game prompt.

Expired50%
Sep 12

The total cost of the entire coding and generation test was only $1.91, whereas a Qwen 3.8 Max test conducted a few weeks earlier cost approximately $32.

Expired50%
Sep 11

In the past two weeks, three new models from China have emerged: Kimi K3, DeepSeek's new Flash model, and Qwen 3.8 Max.

Unverified50%
Sep 11

The daily free token quota for the Qwen 3.8 Max model on the platform is approximately 50 million tokens, having previously reached 10 billion tokens.

Expired50%
Sep 6

The Qwen 3.8 Max 27B version is a traditional dense model, suitable for high-performance desktops or servers.

Unverified60%

All Facts (5)

Source Articles

Qwen 3.8 27B Local Deployment Hands-On: Agent Benchmarks Crush OpusSep 11
通义千问Qwen 3.8 Max实测:开源模型如何逼近顶级闭源水准Jul 23