conv.

All stories
AIQuiet 9d · day 9

PrismML ships Ternary Bonsai 2 27B, a 5.9GB model near Qwen3.8-27B quality

The ternary-quantized model is 9x smaller than its full-precision base while keeping 98.2% of its benchmark performance.

newest coverage 8d agosummary written 8d ago5 pieces691 likes & upvotes44 comments XGoogle News

What to know

  • Ternary Bonsai 2 27B compresses Qwen3.8 27B to 5.9GB, over 9x smaller than the full-precision model.
  • PrismML reports the model retains 98.2% of aggregate performance across a 20-benchmark suite.
  • It is the second Bonsai release, arriving two months after the original Bonsai 27B, with PrismML citing quality as the main improvement.
  • Early reactions from AI accounts on X have been positive, focused on the size-to-quality tradeoff.

PrismML AI lab, developer of Bonsai modelsBabak Hassibi AI researcher / commenterMiaAI_lab AI lab account

PrismML ships Ternary Bonsai 2 27B, a 5.9GB model near Qwen3.8-27B quality
x.com

How it unfolded 4 developments, newest first · click a bar or a number to jump articlesposts

Peak 2 pieces in two hours at Sep 17, 6 PM; 5 pieces over 9 days (1 article · 4 posts) Sep 17, 4 PM — 1 piece · 1 post — X 1Sep 17, 6 PM — 2 pieces · 1 article · 1 post — Google News 1, X 1Sep 17, 8 PM — quietSep 17, 10 PM — 2 pieces · 2 posts — X 2Sep 18, 12 AM — quietSep 18, 2 AM — quietSep 18, 4 AM — quietSep 18, 6 AM — quietSep 18, 8 AM — quietSep 18, 10 AM — quietSep 18, 12 PM — quietSep 18, 2 PM — quietSep 18, 4 PM — quietSep 18, 6 PM — quietSep 18, 8 PM — quietSep 18, 10 PM — quietSep 19, 12 AM — quietSep 19, 2 AM — quietSep 19, 4 AM — quietSep 19, 6 AM — quietSep 19, 8 AM — quietSep 19, 10 AM — quietSep 19, 12 PM — quietSep 19, 2 PM — quietSep 19, 4 PM — quietSep 19, 6 PM — quietSep 19, 8 PM — quietSep 19, 10 PM — quietSep 20, 12 AM — quietSep 20, 2 AM — quietSep 20, 4 AM — quietSep 20, 6 AM — quietSep 20, 8 AM — quietSep 20, 10 AM — quietSep 20, 12 PM — quietSep 20, 2 PM — quietSep 20, 4 PM — quietSep 20, 6 PM — quietSep 20, 8 PM — quietSep 20, 10 PM — quietSep 21, 12 AM — quietSep 21, 2 AM — quietSep 21, 4 AM — quietSep 21, 6 AM — quietSep 21, 8 AM — quietSep 21, 10 AM — quietSep 21, 12 PM — quietSep 21, 2 PM — quietSep 21, 4 PM — quietSep 21, 6 PM — quietSep 21, 8 PM — quietSep 21, 10 PM — quietSep 22, 12 AM — quietSep 22, 2 AM — quietSep 22, 4 AM — quietSep 22, 6 AM — quietSep 22, 8 AM — quietSep 22, 10 AM — quietSep 22, 12 PM — quietSep 22, 2 PM — quietSep 22, 4 PM — quietSep 22, 6 PM — quietSep 22, 8 PM — quietSep 22, 10 PM — quietSep 23, 12 AM — quietSep 23, 2 AM — quietSep 23, 4 AM — quietSep 23, 6 AM — quietSep 23, 8 AM — quietSep 23, 10 AM — quietSep 23, 12 PM — quietSep 23, 2 PM — quietSep 23, 4 PM — quietSep 23, 6 PM — quietSep 23, 8 PM — quietSep 23, 10 PM — quietSep 24, 12 AM — quietSep 24, 2 AM — quietSep 24, 4 AM — quietSep 24, 6 AM — quietSep 24, 8 AM — quietSep 24, 10 AM — quietSep 24, 12 PM — quietSep 24, 2 PM — quietSep 24, 4 PM — quietSep 24, 6 PM — quietSep 24, 8 PM — quietSep 24, 10 PM — quietYesterday, 12 AM — quietYesterday, 2 AM — quietYesterday, 4 AM — quietYesterday, 6 AM — quietYesterday, 8 AM — quietYesterday, 10 AM — quietYesterday, 12 PM — quietYesterday, 2 PM — quietYesterday, 4 PM — quietYesterday, 6 PM — quietYesterday, 8 PM — quietYesterday, 10 PM — quietToday, 12 AM — quietToday, 2 AM — quietToday, 4 AM — quietToday, 6 AM — quietToday, 8 AM — quietToday, 10 AM — quietToday, 12 PM — quietToday, 2 PM — quiet 1–4
Sep 18Sep 19Sep 20Sep 21Sep 22Sep 23Sep 24yesterdaynow · 4:29 PM ET
  1. 4

    Researcher praises narrowing quality gap for ternary models

    Babak Hassibi posted that he was excited about the release and highlighted how quickly ternary models have closed the quality gap with full-precision counterparts.

    “It is remarkable how quickly the quality gap between ternary and full-precision models has been narrowed.”
    — @babakhassibi

    • Very excited about today's release of Bonsai 2 27B. It is remarkable how quickly the quality gap between ternary and full-precision models has been narrowed. Across a suite of 20 benchmarks, Ternary Bonsai 2 27B retains 98.2% of Qwen3.8 27B's aggregate performance, despite being more than 9x smalle...

      @babakhassibiX8d agoview on X ↗
    1 more of the top 2 · 2 posts in this stretch

    • Today, we're announcing Ternary Bonsai 2 27B. Based on Qwen3.8 27B, Bonsai 2 27B is 9x smaller than its full-precision counterpart while retaining 98.2% of its aggregate benchmark performance. Two months after the first Bonsai 27B release, the biggest change is quality. The footprint remains 5.9 ...

      @prismmlX8d agoview on X ↗
    all of them →
  2. 3

    AI lab account reacts, plans to test the model

    MiaAI_lab posted a reaction expressing surprise at the size-to-quality tradeoff and said it would test the model.

    “9 GB but almost as good as Qwen3.8-27B? 😮 I'll test it out!”
    — @MiaAI_lab

    • Ternary Bonsai 2 27B 5.9 GB but almost as good as Qwen3.8-27B? 😮 I'll test it out!

      @MiaAI_labX8d ago15▲view on X ↗
  3. 2

    Coverage details the model's 5.9GB footprint

    GIGAZINE reported on the release, noting the model reduces the size of Qwen3.8 27B to 5.9GB while maintaining 98.2% of its performance.

    1. 1 outlet The 'Ternary Bonsai 2 27B' is an AI model that reduces the size of the Qwen3.8 27B to 5.9GB while maintaining 98.2% of its performance.

      first by GIGAZINE, 8d ago

  4. 1

    PrismML announces Ternary Bonsai 2 27B

    PrismML posted the release of Ternary Bonsai 2 27B, a ternary-quantized model based on Qwen3.8 27B that is 9x smaller in footprint while retaining 98.2% of aggregate benchmark performance, two months after the original Bonsai 27B.

    “Based on Qwen3.8 27B, Bonsai 2 27B is 9x smaller than its full-precision counterpart while retaining 98.2% of its aggregate benchmark performance.”
    — @PrismML

    • Today, we’re announcing Ternary Bonsai 2 27B. Based on Qwen3.8 27B, Bonsai 2 27B is 9x smaller than its full-precision counterpart while retaining 98.2% of its aggregate benchmark performance. Two months after the first Bonsai 27B release, the biggest change is quality. The

      @PrismMLX8d ago676▲view on X ↗

What people are saying 0 voices from 0 sites · best of 4 · verbatim