← All models

GPT-5.1 Thinking

Published Publisher OpenAIgpt-5.1

Overview

GPT-5.1 Thinking is OpenAI’s advanced reasoning model, announced for ChatGPT on November 12, 2025. It adapts its thinking time to the difficulty of a request, responding faster to simpler tasks and spending more time on complex ones. OpenAI announced the corresponding API model as gpt-5.1; the API release followed on November 13, 2025. (openai.com)

Capabilities and uses

The model is intended for complex reasoning, technical explanations, coding and multi-step agentic workflows. The API supports configurable reasoning effort (none, low, medium, and high), function calling and structured outputs. It accepts text and image inputs and produces text output. (developers.openai.com)

Limits and availability

  • The API model has a 400,000-token context window, a maximum output of 128,000 tokens and a knowledge cutoff of September 30, 2024. Audio, video and fine-tuning are not supported. Responses can still contain errors and should be verified where accuracy matters. (developers.openai.com)
  • GPT-5.1 Thinking was retired from ChatGPT on March 11, 2026. The gpt-5.1 API model is deprecated and is scheduled for shutdown on April 1, 2027. (help.openai.com)

API pricing

As checked on November 17, 2026, OpenAI lists standard text-token pricing for gpt-5.1 in USD per 1 million tokens: $1.25 input, $0.125 cached input, and $10.00 output. Cached-input pricing applies to eligible cached tokens; other services or tools may incur separate charges. These are API rates, not ChatGPT subscription prices. (developers.openai.com)

Official sources