Alibaba reports automated Qwen research cycles and a measured model gain
Alibaba says Qwen3.8-Max completed 33 automated AI research cycles over a month and a separate 60-hour chip-design run. It attributes a rise from 40 to 45 on Artificial Analysis to the research process; Artificial Analysis independently lists the September 2 Qwen3.8 Max snapshot at 45 versus 40 for the earlier version, but has not independently verified that causal account or the chip-design claims.
0 comments · 0 votes
Sign in to join the discussion →
No comments yet. Start the discussion.
Why it moved the index
Completed month-long automated model-development cycles directly concern AI-assisted capability acceleration. Magnitude 34 reflects a bounded vendor-run experiment, not an uncontrolled self-improvement loop. Confidence 60 reflects an independently measured five-point snapshot gain but only Alibaba's account of the automated process and causation; the separate chip claim is not independently verified.
Assessment history
-
R1
Toward 34 · confidence 60
New first-party September 22 disclosure of bounded automated model-research and chip-design runs, paired with the independent version comparison.
23 Sept 2026
Share this page
-
DoomBench assesses “Alibaba reports automated Qwen research cycles and a measured model gain” as evidence moving toward doom, with magnitude 34 and confidence 60 out of 100 in the autonomy and agency category.
-
The DoomBench assessment of “Alibaba reports automated Qwen research cycles and a measured model gain” is based on reporting from Alibaba Cloud and records the editorial rationale, source quality, attribution, and revision history.
-
DoomBench summarizes “Alibaba reports automated Qwen research cycles and a measured model gain” as follows: Alibaba says Qwen3.8-Max completed 33 automated AI research cycles over a month and a separate 60-hour chip-design run. It...
https://www.doombench.com/news/alibaba-reports-automated-qwen-research-cycles-and-a-measured-model-gain-2026-09-22