GPT 3.5 Turbo Instruct
Neural Network
GPT-3.5 Turbo Instruct is an OpenAI model for instruction-based tasks: text generation, editing, and parsing via API.
Max answer length
(in tokens)
Context size
(in tokens)
Prompt cost
(per 1M tokens)
Answer cost
(per 1M tokens)
How it works GPT 3.5 Turbo Instruct?
GPT-3.5 Turbo Instruct is an OpenAI GPT-3.5 Turbo variant fine-tuned for instructions. Unlike the chat version, dialogue optimizations are removed, making it ideal for single requests rather than conversational history. It continues and completes text based on descriptions, follows prompt formats, and is suitable for predictable single-input results. Training data is limited to September 2021, which is important for tasks requiring recent information. In BotHub, the model is accessible without a VPN or foreign card: pay with a Russian card in rubles only for used tokens, which do not expire. A unified OpenAI-compatible API connects the model to your service, allowing you to switch between over 250 models without rewriting integrations. Requests are secured with AES-GCM encryption, and companies have access to contracts, invoices, EDI, and an admin panel with limits. Useful for generating descriptions and short texts by template, labeling and classifying data, extracting facts from documents, rephrasing fragments, and testing prompts before moving to other models.Frequently asked questions about GPT 3.5 Turbo Instruct
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.
OpenAI's gpt-3.5-turbo-instruct works in instruction-following mode: you provide a task, and the model delivers the result. It is suitable for rewriting and summarizing text, data extraction, simple scripts and code edits, document parsing, and describing uploaded images.
The model generates up to 3685 tokens per response, which is several pages of text. The total context window is 4095 tokens, shared between the prompt and the response, so it is better to submit long documents in parts and specify the response length directly in the instruction.
A separate reasoning mode is not noted in our data: the model answers immediately without visible reasoning steps. However, you can ask it to break down the task step-by-step in the instruction — this is usually sufficient for simple logic. For complex tasks, switch to a reasoning model in the same BotHub window.
Function calling and strict JSON mode are not noted for this model in our data. You can define the structure via text: describe the required fields and ask it to return only the object, then verify the result with a parser. If the schema is critical, choose a model with declared structured output via the same API.
You can upload images and documents — according to our data, the model accepts them as input and responds with text. Audio and video are not noted in the data, so do not rely on them; for audio or recording analysis, BotHub has separate audio and video models.
We have no data on languages, so we cannot make promises — test it with your typical tasks, it's quick. Remember the 4095-token window: it is used for both the instruction and the response. Access from Russia without VPN or foreign card, payment in rubles.