BGE-M3
The multilingual workhorse — 100+ languages, MIT, three retrieval modes.
90
sovereigntyBAAI's M3 does dense, sparse, and multi-vector retrieval in one MIT-licensed model across 100+ languages — the open pick when your corpus isn't English or you want hybrid search signals without running two systems. At scale on a spot GPU it embeds for roughly $0.001 per million tokens.
OPEN SOURCEMITSELF-HOSTLOCAL-FIRST
LicenseMIT
PricingFree (MIT) — GPU recommended for throughput; ~$0.001/1M tokens at spot-GPU scale
Open sourceYes
Self-hostableYes
Local-first dataYes
What it does well
- +100+ languages in a single model
- +Dense + sparse + multi-vector retrieval built in
- +MIT license with strong community adoption
Where it falls short
- −Heavier to serve than small English models
- −Wants a GPU for production throughput
BGE-M3 as an alternative to
Where BGE-M3 shows up in our comparisons, and how it ranked.
BGE-M3 head-to-head
Straight comparisons against the tools people weigh it against.