GPT OSS 20B Batch
Neural Network
GPT OSS 20B is an open OpenAI model with Mixture-of-Experts architecture in batch mode for mass text processing tasks.
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 Batch?
GPT OSS 20B is an open OpenAI model: 21 billion parameters, Mixture-of-Experts architecture, with 3.6 billion active parameters per pass, requiring fewer computations than a dense model of the same size. The Apache 2.0 license resolves weight usage concerns. The batch variant is designed for batch, not interactive work: requests are queued and returned in ready-made batches. In BotHub, the model is accessible from Russia without a VPN or foreign card, and you pay in rubles per token used, without a subscription, and unused tokens do not expire. A single OpenAI-compatible API provides access to the entire catalog, allowing you to switch models without rewriting integrations; traffic is encrypted via AES-GCM, and for companies, we offer contracts, invoices, EDI, and an admin panel with limits. The batch approach is ideal where volume matters more than instant response: labeling and classifying large text arrays, generating uniform catalog descriptions, summarizing document archives and correspondence, preparing synthetic datasets, offline runs, and prompt comparison on samples.Frequently asked questions about GPT OSS 20B Batch
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 a 131,072 token context and reasoning mode. It is suitable for coding, document analysis, analytics, STEM tasks, and agent scenarios with tool calling. Batch mode is convenient when you need to process many similar requests in a batch.
The output limit is up to 117,964 tokens per response, which easily covers long technical texts, extensive refactoring, detailed analyses, and structured reports. The total dialogue context, including input, is limited to 131,072 tokens.
Yes, reasoning mode is confirmed: the model builds a chain of intermediate steps before the final answer. This significantly helps with math, logic, code debugging, and tasks where multiple conditions must be considered sequentially rather than answering immediately.
Yes, both function calling and structured JSON output are supported. The model can be connected to external tools and services, and via the unified OpenAI-compatible BotHub API, you can switch to another model without rewriting the integration.
According to our data, in addition to text, documents are accepted as input — they are convenient to parse, summarize, and extract data from. For images, audio, and video, BotHub has separate multimodal models available in the same interface.
The model is designed for multilingual work, so it processes Russian-language requests. It is best to evaluate the quality on your own tasks: in BotHub, you can send the same prompt to multiple models in one window and compare the results.