OpenAI has presented GPT-5.1This update is designed to make ChatGPT more agile in its reasoning and more natural in its conversations. In this version, the company combines under-the-hood improvements with a clear shift towards... tone customizationSomething that many users had been requesting for some time.
The novelty comes in the form of two complementary variants They divide the work according to the task, along with adjustments that fine-tune how it follows instructions, retains context, and decides how much "reflection" to apply before responding. For Spain and the rest of Europe, the rollout is progressive and it starts this week on the payment plans.
Two GPT-5.1 with different ways of responding
The model unfolds into GPT-5.1 Instant y GPT-5.1 ThinkingInstant aims for a more approachable and direct style, ideal for daily inquiries and clear formats; Thinking, on the other hand, prioritizes the clarity with less jargon and dedicates more time to complex problems, speeding up the simple ones.
In practice, Instant feels more conversational and warm by defaultWhile Thinking reinforces deep reasoning and persistence when the issue becomes complex. Both improve the following instructions and the quality of the code, with visible progress in technical tests.
If you don't want to choose manually, the selector allows you to activate the mode Carswhich decides on the fly which variant best suits your request. Paid users will see the ChatGPT 5.1 model available and will be able to switch between them. Instant and Thinking from the models menu.
OpenAI emphasizes that these improvements are already noticeable in math and programming challenges, with improvements in benchmarks like AIME 2025 and Codeforces. In other words, there's more common sense when needed and less beating around the bush when it's not.
Adaptive reasoning and performance
One of the key features of GPT-5.1 is that it adjusts in a way dynamics their “time for reflection” before answering. It doesn't dedicate the same effort to everything: it reduces calculation in the easy cases and adds layers of analysis in the complex ones to increase coherence and context.
This adaptive reasoning brings answers faster at simple tasks and more elaborate when the challenge demands it. According to OpenAI, the model makes better use of tokens and reduces overhead, which also benefits automated integrations and API costs.
Beyond speed, the internal logic and consistency in long dialogues have been refined. In scenarios with many iterations, the AI maintains improved continuityavoiding unnecessary repetitions and thread loss.
Control of tone and personalities
GPT-5.1 adds new ways to adjust the wizard's style: there are new tone presets and the ability to define features such as conciseness or warmth from the settings. The idea is to adapt ChatGPT to both professional environments and more informal uses.
OpenAI speaks of a broader set of personalities, with additions such as Professional, Sincere and Uniqueand a range that can include up to eight profiles (for example, Default, Friendly, Efficient, Geeky, or Cynical). In the near future, the chatbot itself may suggest changes style if it detects repeated patterns in your requests.
Memory, context and compliance in Europe
Context retention is more effectiveThis facilitates projects with multiple sessions or large documents. In writing, support, or analysis tasks, the continuity is noticeable and appreciated.
For European companies, GPT-5.1 includes options for disable conversation logging when data protection policy requires it, helping to comply with the GDPR without sacrificing the model's capabilities. This control is especially useful in customer service and internal knowledge.
Availability: Spain and deployment plans
The launch begins first in the payment plans (Plus, Pro, Go, and Business) and will be extended to free accounts later. In Spain, the update begins rolling out this week in stages.
Enterprise and Education customers will have access to a seven-day early accessAfter that period, GPT-5.1 will be established as the default model. API It will also be enabled for developers throughout the week.
During the transition, previous GPT-5 models (Instant and Thinking) will remain available in the “Legacy” menu for three months For paid subscribers, in order to compare changes and adapt calmly. Meanwhile, the mode Cars will direct each query to the most appropriate variant if one is not chosen manually.
With 800 over millions of users According to OpenAI, the strategy involves a ChatGPT that adapts to each user profile: more flexible with tone, more precise with instructions, and better able to alternate between speed and depth when necessary. For those who use the tool daily or at work, the goal is clear: less friction and more control about experience.