Ternary Bonsai 2 2B

Released
2026-10-15
Last refreshed
2026-10-20
Status
Researched 1d ago
Open sourceCommercial use: permittedMultimodalRAGAgentsLong contextVisionJSON / Tool useMultimodalAgents

Ternary Bonsai 2 2B is available now for rag, agents, and long context with open-source and 256K context; evaluate it while provider pricing coverage matures.

PrismML releases · 11 in the last 12 monthsChangelog →
Specifications
Family
Bonsai
Released
2026-10-15
Context
256K
Parameters
1.88B
Specialization
on-device
Openness
Open source
License
Apache 2.0OSI-approvedPermitted
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 pricing is available yet.

About

Ternary Bonsai 2 2B is PrismML's ternary open-weight model derived from Qwen3.5-2B, released October 15, 2026 as part of the Bonsai 2 family. It adds vision input through a separate projector and is tuned for local assistants: instruction following, tool calling on mobile devices and concise answers, with a 262K-token context. The smallest packing is 0.43 GB. Apache 2.0.

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-2-2bprism-ml/Ternary-Bonsai-2-2B-ggufBonsai 2 2B