Google details EmbeddingGemma 2: a 740M-parameter Pareto-optimal omni-modal embedder
tomaarsen · x · 2026-10-28
Google's EmbeddingGemma2 team published its technical report (arXiv 2610.37364) on Hugging Face. At just 740M parameters, it's one of the smallest omni-modal embedding models around, Pareto-optimal across multiple modalities, and supports toggling off parts of the model for an even smaller footprint. Hugging Face's Tom Aarsen calls it one of the most exciting embedding releases in months.
Related event: Google Open-Sources EmbeddingGemma 2, a 744M Omni-Modal Embedding Model(6 posts)→
More from Models
- GitHub's Copilot Fusion matches Opus 5.5 coding agents at half the cost per task — ArtificialAnlys · 2026-10-29
- RSI Arena paper: 10 AI agents autonomously post-train models, 8 improve with human feedback — teortaxesTex · 2026-10-29
- ARC-AGI-3 shows a striking outlier: Argon low jumps from 42.1% to 97.5% — GregKamradt · 2026-10-29
- ARC Prize finds v2 = v1^4.53 power curve explains 98% of ARC-AGI-2 score variance — GregKamradt · 2026-10-29
- Don't get hyped: today's OpenAI news may just be Pro 200 limits not getting cut — Angaisb_ · 2026-10-29
- Cryptographer points abliterated GLM at Telegram's secret-chat crypto — and it 'killed itself' — matthew_d_green · 2026-10-29