Fetching from the wire…
01
02
03
04
05
06
07
08
Skills2026-03-19 · source-backed
The #1 MTEB embedding model supports float8 + PCA combinations achieving 8x compression. A 1TB vector index becomes ~125GB. Binary quantization available for extreme compression. 12.6M tokens/hour at $0.22/1M tokens on ml.g6.xlarge. Voyage AI Blog
Each link below shares sources, entities, or timing with this story.
Voyage uses Matryoshka / Shared entity: Voyage / Shared topic / What happened next
Linked by a graph relationship (Voyage uses Matryoshka); both cover Voyage; overlapping topics (embedding, matryoshka).
Shared entity: MTEB / Shared topic / What happened next / Tension
Both cover MTEB; overlapping topics (embedding, model, mteb); picks up the MTEB thread on 2026-03-21.
SkMTEB uses MTEB / Shared entity: MTEB / What happened next
Linked by a graph relationship (SkMTEB uses MTEB); both cover MTEB; picks up the MTEB thread on 2026-06-13.
Shared entity: Binary / What happened next / Tension
Both cover Binary; picks up the Binary thread on 2026-07-30; pushes against this story (against).
Shared entity: Voyage / What happened next / Tension
Both cover Voyage; picks up the Voyage thread on 2026-06-08; pushes against this story (but).
Voyage uses Matryoshka / Shared topic
Linked by a graph relationship (Voyage uses Matryoshka); overlapping topics (embedding, loss).
Shared topic
Overlapping topics (binary, compression, index, loss, quantization).
Shared entity: Binary / What happened next
Both cover Binary; picks up the Binary thread on 2026-08-17.