Stella 1.5B v5
Independent researcher's 1.5B embedder that ranked #3 on MTEB
About
stella_en_1.5B_v5 is the work of independent researcher Dun Zhang (NovaSearch), trained on top of Alibaba's gte-large-en-v1.5 and gte-Qwen2-1.5B-instruct. At 1.5B parameters it scored a reported MTEB average of 67.82 — rank 3 at the time, ahead of far larger institutional models — with flexible output dimensions from 1024 up to 8192. It is MIT licensed and freely downloadable. The honest context: this is essentially a one-person research project, so documentation, long-term maintenance, and ecosystem integrations are thinner than for BAAI, Alibaba, or Nomic models. But as a compact high-quality English embedder from an independent author, it earned its place on the leaderboard.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
MTEB average 67.82, rank 3 at release (reported in comparison tables)
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 6/10
Scored 6/10 (metrics tier, metrics-v1). Evidence: 1.545B params, MTEB 67.82 (rank 3), MIT license, trained on gte-large-en-v1.5 + gte-Qwen2-1.5B-instruct corroborated in a research survey and the instill-ai model mirror read this session.; Single independent author (Dun Zhang) — thinner maintenance/ecosystem than institutional models — metrics tier.; Exact release day unpinned (whitepaper late December 2024), left null..