LightOn's SlideOn sweeps MTEB, taking top two spots for under-500M retrieval models
tomaarsen · x · 2026-10-22
tomaarsen, maintainer of sentence-transformers, highlights that LightOn's SlideOn and mSlideOn now hold the best and second-best MTEB (Eng, v2) retrieval scores ever for models under 500M parameters. The leaderboard screenshot shows SlideOn (174M) at rank #5 with a 58.62 mean and mSlideOn (339M) at #16 with 58.07, beating larger rivals like Snowflake Arctic Embed and jina-embeddings.
More from Models
- Stanford's Token-Gated LLMs skim context for up to 11x faster long-input prefill — CShorten30 · 2026-10-22
- Secret Anthropic model 'Fable 5.5' builds voxel web app in 35 minutes, no visual feedback — imjustnewatai · 2026-10-22
- Towards AI builds writing benchmark: 4 LLM judges, 1000 human A/B tests on AI slop — AI Engineer · 2026-10-22
- NVIDIA unveils Vera Rubin full-stack agent hardware, claims post-trained Nemotron beats frontier models — mervenoyann · 2026-10-22
- Carbon checkpoint on LMArena updated, Gemini web and app change in sync — lyraxana · 2026-10-22
- KRAFTON's Raon-1-mini-preview Trends on Hugging Face — KRAFTON · 2026-10-22