Fetching from the wire…
01
02
03
04
05
06
07
08
Public story · 2026-08-24 · high
A multi-model study found refinement quality tracks generator and refiner size, but barely moves with critic size at all.
Why now: As of August 24, 2026, this is the clearest data yet on where model size pays off in a self-refinement loop.
Researchers tested self-refinement loops across five benchmarks, using six sizes of Qwen3 and four sizes of Gemma 3 to isolate which stage of a generate-critique-revise loop drives quality most. The pattern held across both model families: bigger generators and bigger refiners reliably produce better final output. Critic size barely mattered.
The instinct when building one of these loops is to assume every stage benefits from a stronger model, so budget gets spread evenly. This self-refinement study says that instinct is wrong for the critique step specifically. A tiny critic and a huge critic land in about the same place, as long as a critic exists at all.
An undersized refiner is riskier than an undersized critic. Skimp on the generator or refiner and results can fall below the unrefined baseline. Skimp on the critic and the loss is close to nothing.
For anyone running a three-stage loop and paying per token, route critique to the cheapest model that still produces a coherent critique, then put the saved budget into the generator or refiner. Skipping critique entirely still underperforms including even a small one. The move is shrinking the critic, not cutting it.
Each link below shares sources, entities, or timing with this story.
Gemma built by Google / Shared entities / Shared topic / Earlier coverage
Linked by a graph relationship (Gemma built by Google); both cover Gemma, Qwen3; overlapping topics (benchmark, gemma, model).
Gemma built by Google / Shared entity: Gemma / Shared topic / Earlier coverage
Linked by a graph relationship (Gemma built by Google); both cover Gemma; overlapping topics (beat, gemma, model).
Linked by a graph relationship (Gemma built by Google); both cover Gemma; overlapping topics (gemma, model).
Linked by a graph relationship (Gemma built by Google); both cover Gemma; overlapping topics (gemma, model).
Gemma built by Google / Shared entity: Gemma / Earlier coverage / Tension
Linked by a graph relationship (Gemma built by Google); both cover Gemma; earlier Gemma coverage from 2026-06-14.
Gemma built by Google / Shared entity: Gemma / Shared topic / Earlier coverage
Linked by a graph relationship (Gemma built by Google); both cover Gemma; overlapping topics (beat, benchmark, gemma, model).
DiffusionGemma benchmarked against Gemma / Shared entity: Gemma / Earlier coverage
Linked by a graph relationship (DiffusionGemma benchmarked against Gemma); both cover Gemma; earlier Gemma coverage from 2026-06-11.
Gemma built by Google / Shared entity: Gemma / Shared topic / Earlier coverage / Tension
Linked by a graph relationship (Gemma built by Google); both cover Gemma; overlapping topics (benchmark, model).