Ternary Bonsai 27B

Released
2026-07-14
Last refreshed
2026-09-17
Status
Researched 25d ago
Open sourceCommercial use: permittedMultimodalRAGAgentsLong contextVisionJSON / Tool useMultimodalAgents

Ternary Bonsai 27B is available now for rag, agents, and long context with open-source and 256K context; evaluate it while provider pricing coverage matures.

PrismML releases · 2 in the last 12 monthsChangelog →
Specifications
Family
Bonsai
Released
2026-07-14
Context
256K
Parameters
27B
Specialization
on-device
Openness
Open source
License
Apache 2.0OSI-approvedCommercial use: permitted
Weights
Available
Code
Unknown
Training
Fine-tuned
Created by

PrismML (Prism ML, Inc.) builds ultra-dense 1-bit and ternary Bonsai models for on-device and efficient datacenter infer

Website
Pricing

No tracked provider token pricing is available yet.

About

Ternary Bonsai 27B is PrismML's prior-generation ternary 27B open-weight multimodal model from the 2026-07-14 Bonsai 27B launch (MCP-verified @PrismML 2077084891284721827 + first-party news/blog). Based on Qwen3.6 27B (live catalog slug qwen3.6-27b). Announce: ~5.9 GB footprint, laptop-class ternary variant alongside 1-bit Bonsai 27B (~3.9 GB phone-class. Apache 2.0; weights on HF prism-ml/Ternary-Bonsai-27B-gguf (+ mlx-2bit). Tip Bonsai 2 is the follow-up ~2 months later under the same Bonsai family. Architecture notes: Ternary (~1.58–1.71 bpw) end-to-end language weights; multimodal with separate vision tower

Providers

No active provider routes are tracked for this model yet.

Capabilities

VisionMultimodalReasoningJSON / Tool use

Benchmark peer barsfor RAG

No task-mapped benchmark peers are available for this model yet.

Migration checks

No linked migration route is available for this model yet.

API versions

ternary-bonsai-27b