cal/claude-memory
1
0
Fork 0
Code Issues 2 Pull Requests Actions Packages Projects Releases Wiki Activity

store: Cloud embedding APIs offer no advantage over local qwen3-embedding:8b

Browse Source
This commit is contained in:
Cal Corum 2026-02-19 14:53:23 -06:00
parent 99114ea561
commit b3023a9bf7
1 changed files with 12 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

12
graph/insights/cloud-embedding-apis-offer-no-advantage-over-local-qwen3-emb-97866f.md Normal file
View File

@ -0,0 +1,12 @@
---
id: 97866ffd-f148-4dda-b4c6-7cd9c7861b63
type: insight
title: "Cloud embedding APIs offer no advantage over local qwen3-embedding:8b"
tags: [embedding, privacy, openai, ollama, comparison]
importance: 0.6
confidence: 0.8
created: "2026-02-19T20:53:23.330619+00:00"
updated: "2026-02-19T20:53:23.330619+00:00"
---
Researched all major cloud embedding APIs (Feb 2026): OpenAI text-embedding-3-large (64.6 MTEB, $0.065-0.13/1M tokens), Google gemini-embedding-001 (68.3, $0.15/1M), Voyage AI voyage-3.5 (66-67, $0.06/1M), Cohere embed-v4 (65.2, ~$0.10/1M). Local qwen3-embedding:8b scores ~70.6 MTEB — beats all of them. Privacy comparison: OpenAI retains 30 days; Voyage AI trains on data by default with irrevocable license (!); Google paid API doesn't train; Cohere opt-in. Local = free + fully private + highest quality. No reason to use cloud for this use case.