OpenAI GPT-3.5 Turbo Instruct

GPT-3.5 Turbo Instruct is the odd member of the GPT-3.5 family. Its siblings were chat models; this one was trained to take a single instruction and continue the text, the way the older GPT-3 models worked. OpenAI kept it in the catalogue for customers whose code had been written against the completions endpoint and could not be rewritten as a conversation. It is also the family member that aged worst on price. It costs $1.50 per million input tokens and $2.00 per million output tokens, against $0.50 and $1.50 for the chat model gpt-3.5-turbo. That is three times as much on input for a model with a quarter of the context window: 4,096 tokens against 16,385. It cannot be fine-tuned, offers no tool calling and accepts text only. Its knowledge stops in September 2021. OpenAI deprecated it on 26 September 2025 together with gpt-3.5-turbo-1106 and shuts it down on 28 September 2026, naming gpt-5.6-terra as the replacement. The chat model it was built alongside outlives it by under a month, to 23 October 2026. When both are gone, so is the last part of the GPT-3.5 generation.

▸News

The GPT-3.5 that OpenAI switches off on 28 September charged three times the chat model's input price for a quarter of the context

9/16/2026 · OpenAI, model deprecations and model cards

▸Videos

No videos yet.

▸Reviews