Ternary Bonsai 2 27B
Neural Network
Ternary Bonsai 2 27B is a PrismML reasoning model with a 262K context window for code, math, and images.
Specifications
- Input
- text, images, documents
- Output
- text
- Context size
- 262 144 tokens
- Max answer length
- 32 768 tokens
- Reasoning effort
- none, medium, extra high
- Capabilities
- reasoning, function calling
Prompt cost
(per 1M tokens)
Answer cost
(per 1M tokens)
How it works Ternary Bonsai 2 27B?
Bonsai 2 27B is a reasoning model by PrismML with 27 billion parameters, based on Qwen3.8-27B. It excels in programming and mathematics, supports function calling, and understands images. Its 262K context window allows handling large repositories, long technical documents, and multi-step dialogues; it is based on ternary compression, which gives the model its name. Via BotHub, it is accessible from Russia without a VPN or foreign card; payment is in rubles per token used, with no subscription, and unused tokens do not expire. Over 250 models are available in one window for easy comparison, and a unified OpenAI-compatible API allows you to connect the model to your product and switch it without rewriting integrations. Traffic is encrypted via AES-GCM, chats are not saved, and companies have access to contracts, invoices, electronic document management, and an admin panel with limits. Developers will find it useful for code review and refactoring in large projects, engineers for step-by-step calculation analysis, analysts for reading diagrams and report screenshots, and product teams for building agents with external function calls.Frequently asked questions about Ternary Bonsai 2 27B
You can use generated results for commercial purposes. You own all rights to the content you create. The only restriction: make sure your prompt does not include copyrighted third-party material. You are responsible for respecting the rights to any input data.
It is a text model with a 262,144 token context window: it analyzes long documents, writes and refactors code, helps with debugging, answers STEM questions, and reasons before answering. It accepts text, documents, and images as input.
The limit for a single response is 32,768 tokens, which is roughly a large article or an entire large code file. If you need a larger result, break the task into parts: the 262,144 token context allows you to keep the entire conversation and already generated fragments.
Yes, the reasoning mode is confirmed in our data. The model can break down a task into steps before the final answer — this significantly helps with math, logic, code analysis, and multi-step tasks where output accuracy is more important than speed.
Yes, the model supports function calling and structured JSON output. You can connect external tools, databases, and your own services, receiving answers in a predictable schema — this is convenient for agents and pipelines where the result goes directly into code.
Images and documents — yes, they can be sent along with text: the model will analyze a screenshot, diagram, chart, or multi-page file and respond with text. Audio and video are not noted in our data, so it is better to choose a specialized model from the catalog for them.
We do not have data on languages, so we will not make any promises — it is easiest to check in practice. Send your typical request and compare the result with other models in the same window: switching takes a couple of seconds, and payment is per token used.