DeepSeek V3.1 Terminus
Neural Network
DeepSeek V3.1 Terminus is an update to DeepSeek V3.1 with improved language consistency and agent capabilities.
Max answer length
(in tokens)
Context size
(in tokens)
Prompt cost
(per 1M tokens)
Answer cost
(per 1M tokens)
How it works DeepSeek V3.1 Terminus?
Terminus is an update to the DeepSeek V3.1 model from the DeepSeek team: the core capabilities of the previous version are preserved, while some user feedback has been addressed—language consistency and behavior in agent scenarios have been improved. If you have already built processes on DeepSeek V3.1, the transition does not require rebuilding logic: it is the same line, just more refined in the details. In BotHub, the model is available from Russia without a VPN or foreign card; payment is made with Russian cards in rubles and charged per token, and unused balance does not expire. Nearby in the same window are over 250 other neural networks, so you can compare answers and choose the model for your task without creating separate accounts. For product teams, there is a unified OpenAI-compatible API: one connection, and the model behind it can be changed without rewriting the integration, plus AES-GCM encryption and corporate access via contract with an invoice and admin panel. Developers will find Terminus useful for code analysis and refactoring, analysts for compiling reports, support teams for drafting responses, and those building agents for use as a core model.Frequently asked questions about DeepSeek V3.1 Terminus
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 DeepSeek text model with reasoning and tool calling. It is suitable for working with code: writing, refactoring, debugging, as well as analyzing long documents, analytics, and STEM tasks. The context window is 131,072 tokens, so a large project or a package of files will fit into a single dialogue.
Up to 65,536 tokens per response — this is dozens of pages of text or a large code module in its entirety. The model provides long materials, documentation, and refactoring of a large file without splitting it into parts, so you won't have to assemble the result from pieces.
Yes, reasoning before answering and chain-of-thought are confirmed by our data. The model first breaks down the condition step-by-step, checks intermediate conclusions, and only then answers. This significantly helps with mathematics, logic, requirement analysis, and finding the cause of errors in code.
Yes, both. The model calls external functions and returns structured JSON, making it easy to integrate into agents, pipelines, and business logic. In BotHub, this works via a unified OpenAI-compatible API: you can switch models without rewriting the integration.
In addition to text, you can provide images and documents as input: the model will analyze a screenshot, diagram, table, or file and respond with text. Audio and video are not noted in our data. Generation is also text-based — there are separate models in the catalog for images and video.
The language of prompts and responses is not specifically noted in our data, so we will not promise accuracy — it is more reliable to test it on your own task. In the BotHub window, you can compare this model's response with others and choose the right one, without a VPN or foreign card.