cd /news/large-language-models/ternary-bonsai-2-27b · home › topics › large-language-models › article
[ARTICLE · art-135142] src=tokenstead.ai ↗ pub= topic=large-language-models verified=true sentiment=· neutral

Ternary Bonsai 2 27B

Ternary Bonsai 2 27B, a ternary-weight multimodal model built on a Qwen3.8 27B base, ships at roughly 6GB with weights restricted to {-1, 0, +1} at 1.76 bits per weight with FP16 group scales, according to the model's provider. The 27.4B-parameter model offers hybrid attention, a 262k context window, and vision input, and retains 98.2 percent of the FP16 baseline on a 20-benchmark thinking suite while running on llama.cpp and MLX. The provider reports the model adds no embedded watermarks or provenance metadata to generated output, and no per-token API provider pricing is tracked for it yet.

read1 min views1 publishedSep 20, 2026
Ternary Bonsai 2 27B
Image: Tokenstead (auto-discovered)

consumer Ternary-weight multimodal model on a Qwen3.8 27B base: weights are values in {-1, 0, +1} at 1.76 bits per weight with FP16 group scales, so the full 27B model ships at roughly 6GB. Hybrid attention, 262k context, vision input. Keeps 98.2 percent of the FP16 baseline on a 20-benchmark thinking suite. Runs on llama.cpp and MLX.

AI-generated content marks

The provider reports that this model does not add embedded watermarks or provenance metadata to generated output.

  • 27.4B
  • 262k
  • apache 2.0
  • Undisclosed
  • Sep 2026

Save your hardware and every model page answers the real question: will it run on your machine, and how fast?

Join free - save your rig →

Or run it in the cloud #

    No per-token API provider pricing tracked for Ternary Bonsai 2 27B yet.

        For flagship list prices, see the
        [calculator](https://tokenstead.ai/calculator).

Inference cost over time #

Data accumulates from the first daily sync - longer ranges populate over time. Prices come from OpenRouter snapshots, not a historical API.

source & further reading
tokenstead.ai — original article Jeffrey Guenther Rack Cellar Typecraft's homelab NVIDIA RTX PRO 5500 Blackwell 84GB
── more in #large-language-models 4 stories · sorted by recency
── more on @ternary bonsai 2 27b 3 stories trending now
sponsored brought to you by zahid.host 4,200+ EU-deployed projects
reading about agents? ship yours in a single git push.

Run your AI side-project on zahid.host

EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain — perfect for shipping the agent you just read about.

$git push zahid main
→ Live at https://your-agent.zahid.host ✓
Get free account → Pricing
from €0/mo · no card required
LIVE [news/ternary-bonsai-2-27b] indexed:0 read:1min 2026-09-20 · —