Development

A 35B model beats its own 235B predecessor

development2026-02-25

Claim

Alibaba's Qwen 3.5 at 35 billion parameters surpassed its own 235-billion predecessor, while Inception Labs claimed Mercury 2 is the fastest reasoning model by replacing autoregression with diffusion.

Reported metric: 35B > 235B

Reporting date: 2026-02-25 (not necessarily the event date)

Sources

Attribution and reporting linked by this record; links do not imply independent verification.

Welcome to February 25, 2026 (theinnermostloop.substack.com)

Knowledge graphopen in browser →

About

Mercury 2

Reported in

Welcome to February 25, 2026

Actor

Evidence for

Supersedes

Two hours of video fits in a million-token window

Referenced by

Cite or correct this record

Citation formatCitation text

A 35B model beats its own 235B predecessor
Innermost Loop KB — secondary record
Reporting date: 2026-02-25
KB permalink: https://nicholsn.github.io/innermost-loop-kb/developments/2026-02-25-qwen-35b-beats-its-own-235b
Claim: Alibaba's Qwen 3.5 at 35 billion parameters surpassed its own 235-billion predecessor, while Inception Labs claimed Mercury 2 is the fastest reasoning model by replacing autoregression with diffusion.
Attributed sources:
- Welcome to February 25, 2026 — https://theinnermostloop.substack.com/p/welcome-to-february-25-2026
This citation does not imply independent verification of linked-source claims.

Suggest a correction — opens a GitHub draft for you to review and submit.