Lara Mini Base achieves multilingual parity with 0.6B active parameters
A new open model trained on 6 trillion tokens across 45 languages, including all 24 EU official languages, outperforms larger models on multilingual and math benchmarks while maintaining single-GPU inference speed.
Researchers have released Lara Mini Base, an open-source language model that achieves competitive performance across English, multilingual, math, and code benchmarks using only 0.6B active parameters. The model was trained from scratch on 6 trillion tokens across English and 44 other languages, cove...
Sign in to read the full analysis
Free account. Full analysis on LLM unit economics, plus the weekly Cost-of-Inference column.
Try it on your own context
You just read the writeup. Now run the thing. Paste a doc or some verbose tool output and watch it shrink — free, no signup.
- Source type
- Primary publication (lab/vendor blog) — our analysis + implication
- Source link
- r/arxiv-compression
- Published
- UTC
- Byline
- By the gotcontext.ai team (editorial standards)
- Correction?
- corrections@gotcontext.ai