@GoSailGlobal: Today, August 3, Alibaba officially released Qwen3.8. With 2.4 trillion parameters and 1M context, coding and office capabilities have improved dramatically in this version, pushing it into the global top tier overall. The preview version ran for two weeks, and today it officially graduated. I immediately put it through a hardcore real-world test—let me lay out the results and process. I came up with a pretty tough challenge: single-file HTML hand...
Summary
Alibaba officially released Qwen3.8, with 2.4 trillion parameters and 1M context. Coding and office capabilities have been significantly improved, entering the global top tier. The author conducted a hardcore real-world test using a single-file HTML N-body simulation.
View Cached Full Text
Cached at: 08/03/26, 11:41 AM
Today, August 3rd, Alibaba officially released Qwen3.8.
With 2.4 trillion parameters and a 1M context window, this version brings a major boost to coding and office productivity, pushing it firmly into the global top tier.
The preview build ran for two weeks, and today it officially graduates.
I immediately put it through a hardcore benchmark test. Let me lay out the results and the process up front.
I came up with a pretty brutal task: hand-code a “galaxy collision” N-body simulation in a single HTML file.
6000+ particles, dual spiral arms, real gravitational physics, tidal tails, plus drag, zoom, playback, and a Chinese data panel.
I had four models each write their own version of the same task, then ran them 2×2 side-by-side on one screen.
What makes this task nasty is that it’s a composite job—physics plus rendering plus UI plus interaction—and flaws only surface under extended runtime.
A pretty screenshot doesn’t mean the simulation is actually correct.
Similar Articles
@Xudong07452910: A hot comment section on Hacker News: Qwen 3.6 27B is the ideal choice for local development. Key findings: dense parameter model, native support for 256k context, running Q8_0 quantized version at 30 tokens/…
Qwen 3.6 27B is a dense 27B model that achieves impressive performance on local hardware with 256k context, running at 30 tokens/s on MacBook Max M5 and 50 tokens/s on RTX 5090, and is considered by some as the first local model with true general intelligence.
@RookieRicardoR: Domestic models break through again, matching top models like Claude 4.6 and Gemini 3.1 Pro. Just tested Qwen3.7-Max, sharing some real thoughts. Last night I topped up as soon as the API went live and chose three tasks (see video) to test Qwen3.7-Max's frontend capabilities…
The user tested Qwen3.7-Max and believes it matches top models like Claude 4.6 and Gemini 3.1 Pro in frontend, computing power, and Agent capabilities. Its reasoning ability has significantly improved, and with monthly iteration speed, it has become a first-tier domestic model.
Qwen3.7: The Agent Frontier (15 minute read)
Alibaba's Qwen team has released Qwen3.7-Max, a proprietary agent-foundation model achieving top scores on multiple benchmarks including Terminal-Bench 2.0, SWE-Pro, and GPQA Diamond, with consistent performance across various code environments.
@YRSM_Simon: Qwen3.8-27B 终于要来了，A3B呢？
阿里云Qwen宣布下周开源Qwen3.8-Max和Qwen3.8-27B的权重，其中Max为2.4T参数模型，主打编码与协作。
@10xmylife: Qwen 3.8-Max official version released. The tools demonstrated in the video are Qwen Office and Qwen Code.
Qwen 3.8-Max official version released. The video demonstrates the Qwen Office and Qwen Code tools, focusing on programming and collaborative office capabilities.