GPT OSS 20B
Neural Network
GPT OSS 20B is an OpenAI open model with Mixture-of-Experts architecture for text tasks, available via the unified BotHub API.
Max answer length
(in tokens)
Context size
(in tokens)
Prompt cost
(per 1M tokens)
Answer cost
(per 1M tokens)
How it works GPT OSS 20B?
GPT OSS 20B is an open model from OpenAI: 21B parameters, Mixture-of-Experts architecture with about 3.6B active parameters per pass, and an Apache 2.0 license allowing free use of weights. This sparse scheme makes the model computationally efficient while retaining its knowledge base. In BotHub, GPT OSS 20B works without a VPN or foreign card: pay with Russian cards in rubles for tokens as you go, with no expiration. Access over 250 neural networks in one window to easily compare answers. The unified OpenAI-compatible API lets you switch models without rewriting integrations. Chats are encrypted, and companies get contracts, invoices, EDI, and an admin panel with limits. It is typically used for writing and editing text, analyzing and commenting on code, summarizing long documents, building assistants via API, and evaluating the open model before self-hosting.Frequently asked questions about GPT OSS 20B
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 an open-weights text model from OpenAI. Suitable for Q&A, writing and analyzing code, refactoring and debugging, document processing, and reasoning tasks. It accepts text, images, and documents as input, and a 131,072-token context allows handling large materials.
Up to 32,768 tokens per response — enough for detailed document analysis, lengthy text, or large code modules. If the material doesn't fit in one response, ask it to continue: previous parts remain in the conversation context.
Yes, the model has a reasoning mode: it breaks down tasks into steps before the final answer. This is noticeable in logic, math, and code analysis, where a sequence of conclusions is important, not just a quick response.
Yes. The model can call functions and provide structured JSON output, making it easy to integrate into agents, pipelines, and custom services. In BotHub, this is done via a unified OpenAI-compatible API: you can switch models without rewriting integrations.
You can upload images and documents — the model accepts them as input and responds with text. Audio and video input is not specified in our data, so it is better to choose a specialized model for such tasks: in BotHub, all of them are available in one window.
There is no specific data on prompt and response languages, so we cannot promise specific quality — test the model on your task. Access from Russia is available without a VPN or foreign card, payment is in rubles, and you can switch to another model in the same window if needed.