●FLASH — Gemini 3.5 Flash is now generally available, billed as the most intelligent model for agentic and coding tasks●TIER — New tiers like 3.1 Pro and 3.1 Flash-Lite are rolling into apps, cloud products, and business tools●PIXEL — The June Pixel Drop adds Gemini music generation, AI video and music creation, and screen-recording reactions●OMNI — Gemini Omni (creation), 3 Deep Think (reasoning), and Deep Research (automation) all advance in parallel●LIVE — Gemini Live's real-time interaction is expanding across Android, Search, YouTube, and connected Google apps●ULTRA — Google AI Ultra offers top model access, Deep Research, Veo 3 video, and a 1M-token context window●FLASH — Gemini 3.5 Flash is now generally available, billed as the most intelligent model for agentic and coding tasks●TIER — New tiers like 3.1 Pro and 3.1 Flash-Lite are rolling into apps, cloud products, and business tools●PIXEL — The June Pixel Drop adds Gemini music generation, AI video and music creation, and screen-recording reactions●OMNI — Gemini Omni (creation), 3 Deep Think (reasoning), and Deep Research (automation) all advance in parallel●LIVE — Gemini Live's real-time interaction is expanding across Android, Search, YouTube, and connected Google apps●ULTRA — Google AI Ultra offers top model access, Deep Research, Veo 3 video, and a 1M-token context window

TAG

hnsw

1 articles

← Back to all tags

gemini-api¹ pgvector¹ semantic-search¹ embeddings¹ postgresql¹ production¹

◈ Gemini API/2026-06-19Advanced

When Your pgvector Search Quietly Gets Worse — Field Notes on Protecting Recall with Gemini Embeddings

A semantic search built on Gemini Embeddings and PostgreSQL pgvector tends to lose precision over months without throwing a single error. These are field notes on the real causes — model pinning, operator/index mismatch, HNSW reindexing, and recall collapse under filters — with working code.