Wire flash
Thom Wolf: 18 of 20 best open models are Mixture-of-Experts
Editorial responsibility
- No named human review is recorded for this page.
- Source reporting is collected, normalized, translated or condensed automatically when needed.
- Automatically published source-backed update
Thom Wolf stated in a post on X that 18 of the 20 best open models today are Mixture-of-Experts, adding that dense models are "basically over." The post frames Mixture-of-Experts architecture as dominant among top-performing open models and suggests dense architectures are being displaced. The text is truncated and does not elaborate on the remaining two models, the criteria used to rank the 20 best open models, or the training know-how referenced in the cut-off sentence. No specific models, benchmarks, dates, or organizations are named in the available text.
Source report
18 of the 20 best open models today are Mixture-of-Experts (MoE), according to Thomas Wolf (@Thom_Wolf), who states that "dense is basically over."
Wolf notes, however, that the know-how to train a good dense model remains relevant — the full point is cut off in the original post.
Source
QuanquanGuNeutral / independent
Part of this Story
Thom Wolf Says 18 of 20 Best Open Models Are Mixture-of-Experts