Dots 3 Note Preview
Neural Network
Dots 3 Note Preview is an open MoE model by Dots Studio: 16B active parameters out of 280B, available via API in BotHub.
Specifications
- Input
- text, images, documents
- Output
- text
- Context size
- 512 000 tokens
- Max answer length
- 460 800 tokens
- Reasoning effort
- low, medium, high
- Capabilities
- reasoning, function calling
How it works Dots 3 Note Preview?
Dots 3 Note Preview is an open-weights text model from Dots Studio, built using a mixture of experts architecture: out of 280 billion parameters, 16 billion are active at each step. It is the lightest model in the Dots 3 family: it requires fewer computations per request than the larger versions in the line. The 'Preview' prefix indicates a preliminary version. In BotHub, the model is available without a VPN or foreign card: you pay with a Russian card in rubles and only for the tokens actually consumed, and the remaining balance does not expire. Nearby in the same interface are over 250 neural networks: you can easily switch between them in one window and compare answers to the same request, and a unified OpenAI-compatible API allows you to change the model in a multi-model service without rewriting the integration. Chats are encrypted with AES-GCM, and data is not saved. The model is useful for running mass text tasks with lower computational costs, building an assistant prototype on open weights, getting acquainted with the Dots 3 line before choosing a larger model, connecting a backup generation route in your product, and comparing response quality with other models on your scenarios.Frequently asked questions about Dots 3 Note Preview
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.
This is a text model with reasoning capabilities: it analyzes long documents, works with code, explains, and structures information. It accepts not only text but also images and documents as input, and a 512,000-token context allows you to work with entire sets of materials.
The limit for a single response is up to 460,800 tokens, which is a lot by standard task measures. You can request a voluminous analysis, a large code fragment, or a detailed report in its entirety without breaking the request into parts or manually stitching pieces together.
Yes, reasoning before answering is supported by the model: it first processes the task step-by-step and then provides the result. This helps with math, logic, code analysis, and multi-step analytics, where not only the final result but also the correct path to it is important.
Yes, function calling and structured JSON output are supported. The model can be connected to your tools, search, or database to receive responses in a strict schema. In BotHub, this works via a unified OpenAI-compatible API, so you won't have to rewrite the integration for every model.
Images and documents — yes, they can be sent along with a text request: the model will analyze a screenshot, diagram, table, or multi-page file. Audio and video input is not noted in our data, so for such formats, it is more convenient to switch to a specialized model in the same window.
Prompt and response languages are not described in our data, so we won't make specific promises — check it with your own task. In BotHub, this takes a couple of minutes: access without VPN or a foreign card, payment in rubles, and you can compare another model right next to it.