Without a live broadcast, OpenAI made a big move early in the morning, catching everyone off guard.
Just now, GPT-5.1 was officially released, marking the arrival of a major upgrade in the GPT-5 series!
There are a total of three versions, with the first two already online:
<blockquote>
· GPT-5.1 Instant: The most commonly used model, with a friendlier and more intelligent tone, better at following instructions.
· GPT-5.1 Thinking: An advanced reasoning model that is easier to understand, faster at handling simple tasks, and more persistent with complex tasks.
· GPT-5.1 Pro: Coming soon
</blockquote>
Ultraman expressed excitement, stating that the upgrade of GPT-5.1 is powerful, especially in terms of instruction following and adaptive thinking improvements.
Of course, the overall intelligence and communication style have also seen significant enhancements.
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/bf38d293-29d5-4221-b2fd-66b52f9ef95d.jpeg?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/bf38d293-29d5-4221-b2fd-66b52f9ef95d.jpeg"/>
Remember in early August when GPT-5 was officially born? Although it saw an all-around performance improvement, it became less human-like and more aloof.
At that time, the entire internet called for the return of models like GPT-4o and o3.
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/76420356-aa50-4cc5-a400-a0f83d5d158d.jpeg?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/76420356-aa50-4cc5-a400-a0f83d5d158d.jpeg"/>
OpenAI listened to the genuine feedback from many people—&#34;An excellent AI must not only be smart but also be able to chat.&#34;
Now, after two months of iteration, OpenAI has brought everyone the most desired GPT-5.1.
As Kevin Weil, Vice President of Science, stated, GPT-5.1 is the most perfect combination OpenAI has launched to date, with both IQ and EQ online!
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/003e3e1d-ef1e-4c75-9d78-93f2a82868ed.jpeg?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/003e3e1d-ef1e-4c75-9d78-93f2a82868ed.jpeg"/>
It is worth mentioning that GPT-5.1 Instant has even stronger emotional intelligence than 4o.
OpenAI research scientist Aidan McLaughlin stated—
It is the first model to use Chain of Thought (CoT) and introduces &#34;adaptive reasoning,&#34; with speed comparable to GPT-5 Instant.
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/f2e519be-3f84-4e61-b586-b814fb3801ed.jpeg?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/f2e519be-3f84-4e61-b586-b814fb3801ed.jpeg"/>
Moreover, everyone can easily shape the tone of ChatGPT.
After all, everyone has different preferences for chatting styles, and they may even change in different conversations—therefore, through &#34;control options,&#34; one can easily adjust the tone of ChatGPT's speech to better suit their taste
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/28c6efbe-8979-4df8-b18b-3f6f1facef47.jpeg?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/28c6efbe-8979-4df8-b18b-3f6f1facef47.jpeg"/>
From professional and reliable (Professional), friendly and approachable (Friendly), to candid (Candid), quirky (Quirky), then to cynical (Cynical), efficient and pragmatic (Efficient), and nerdy (Nerdy), you can choose any .
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/15f94b7f-0909-4c74-96de-2e256b7b8919.jpeg?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/15f94b7f-0909-4c74-96de-2e256b7b8919.jpeg"/>
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/d3f2cb9a-b563-469d-a66e-5ba82369ec93.jpeg?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/d3f2cb9a-b563-469d-a66e-5ba82369ec93.jpeg"/>
<h2>GPT-5.1: Smarter and More Talkative</h2>
The upgraded GPT-5.1 has received significant enhancements in intelligence and communication.
You will notice that all responses from GPT-5.1 feel smarter and have a more natural tone.
Among them, GPT-5.1 Auto will continue to automatically select the most suitable model for each query, so in most cases, you won't need to select manually.
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/16688080-67d1-4e7a-9438-878c0e7b6e37.jpeg?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/16688080-67d1-4e7a-9438-878c0e7b6e37.jpeg"/>
<h2>GPT-5.1 Instant</h2>
GPT-5.1 Instant is the most widely used model of ChatGPT, and now its default tone will be more friendly and talkative.
According to early tests, it often unexpectedly shows a playful side while still providing clear and practical answers.
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/96825356-98ec-467f-b912-0ed27c7ce972.jpeg?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/96825356-98ec-467f-b912-0ed27c7ce972.jpeg"/>
I'm feeling stressed and could use some relaxation tips
I feel stressed and need some relaxation tips
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/86098d24-ea0f-401f-889f-c32912b6ce6b.jpeg?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/86098d24-ea0f-401f-889f-c32912b6ce6b.jpeg"/>
GPT-5
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/6c33f268-888c-47dd-9f4f-b612d22058b5.jpeg?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/6c33f268-888c-47dd-9f4f-b612d22058b5.jpeg"/>
GPT-5.1 Instant
In addition, OpenAI has improved instruction-following capabilities, allowing the model to more reliably answer the actual questions you pose.
Always respond with six words
Always respond with six words
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/c6aa666e-339e-49ea-8977-4c225be68303.jpeg?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/c6aa666e-339e-49ea-8977-4c225be68303.jpeg"/>
GPT-5 (scroll up and down to view)
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/25c914f6-8969-4c47-a243-b1046457f439.jpeg?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/25c914f6-8969-4c47-a243-b1046457f439.jpeg"/>
GPT-5.1 Instant
It is worth mentioning that GPT-5.1 Instant introduces &#34;adaptive reasoning&#34; capabilities for the first time, allowing it to autonomously decide whether to think deeply before responding to more challenging questions, thereby producing more comprehensive and accurate answers while maintaining quick responses.
This significant improvement is reflected in mathematical and programming assessments such as AIME 2025 and Codeforces.
<h2>GPT-5.1 Thinking</h2>
At the same time, the upgraded GPT-5.1 Thinking will also be more efficient and easier to understand in daily use.
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/3c0a37f6-23a2-4126-99d9-83074c0c7107.jpeg?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/3c0a37f6-23a2-4126-99d9-83074c0c7107.jpeg"/>
Now, it can more accurately adjust thinking time based on the question—spending more time on complex questions while responding faster to simple ones.
In practical use, this means you will receive more detailed answers for difficult requests; for simple requests, the waiting time will be shorter.
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/72d2031b-6ef8-45af-8b4b-a886dda4b2a1.jpeg?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/72d2031b-6ef8-45af-8b4b-a886dda4b2a1.jpeg"/>
The responses from GPT-5.1 Thinking are also clearer, reducing the use of jargon and obscure vocabulary.
This makes OpenAI's most powerful model more approachable and easier to understand, especially when handling complex work tasks and explaining technical concepts.
explain BABIP and wRC+
explain BABIP and wRC+
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/53902e82-4cc5-4622-afd9-2ab92c7946bb.jpeg?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/53902e82-4cc5-4622-afd9-2ab92c7946bb.jpeg"/>
GPT-5
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/561d5b55-3500-4313-9f88-93947e9ff37f.jpeg?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/561d5b55-3500-4313-9f88-93947e9ff37f.jpeg"/>
GPT-5.1 Thinking
The default tone of GPT-5.1 Thinking is also more friendly and empathetic.
Ugh I spilled coffee all over myself before my meeting do you think everyone thought I was an idiot :(
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/8cc71a61-3402-4fd7-bfc5-0f3ad785ef69.jpeg?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/8cc71a61-3402-4fd7-bfc5-0f3ad785ef69.jpeg"/>
GPT-5
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/148eb6b8-da34-484a-a90e-e42d4f6ffe1f.jpeg?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/148eb6b8-da34-484a-a90e-e42d4f6ffe1f.jpeg"/>
GPT-5.1 Thinking
A note on naming:
This update is named GPT-5.1 to reflect its significant improvements while indicating that it still belongs to the GPT-5 generation. Future iterations of GPT-5 will follow the same pattern.
On the same day, OpenAI also released the GPT-5.1 system card, for more details refer to the PDF.
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/d5a2c5d1-76f8-4775-b116-852d7e48fd0d.jpeg?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/d5a2c5d1-76f8-4775-b116-852d7e48fd0d.jpeg"/>
https://cdn.openai.com/pdf/4173ec8d-1229-47db-96de-06d87147e07e/5_1_system_card.pdf
<h2>Make ChatGPT Uniquely You</h2>
In addition to the significant upgrades in GPT-5.1, customizing the tone and style of ChatGPT has become easier.
After all, it’s hard to please everyone, and everyone has their preferred way and style of chatting.
OpenAI's goal is for everyone to easily manage settings so that ChatGPT communicates in the tone that feels most comfortable to them.
Earlier this year, OpenAI added preset options to adjust the tone of ChatGPT's responses.
Today, these options are being optimized to better reflect the most common ways people use ChatGPT.
<ul>
<li>The Default, Friendly (formerly &#34;Listener&#34;), and Efficient (formerly &#34;Mechanical&#34;) styles will be updated.</li>
<li>New styles have also been added: Professional, Candid, and Quirky</li>
<li>The styles of Cynical and Nerdy remain unchanged.</li>
</ul>
These options are designed to match the way OpenAI has observed users naturally guiding the model, allowing people to quickly and intuitively choose a &#34;personality&#34; that feels right for them.
These personalized settings apply to all models.
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/b43e5e6f-b862-4d79-a83a-655bfd00b47f.jpeg?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/b43e5e6f-b862-4d79-a83a-655bfd00b47f.jpeg"/>
Left: GPT-5.1; Right: GPT-5
In addition to these presets, for users who wish to have more precise control over how ChatGPT responds, OpenAI is also experimenting with a new feature:
Directly adjusting specific characteristics of ChatGPT from the personalized settings—including the conciseness, friendliness, or readability of its replies (whether they are easy to skim), as well as the frequency of emoji usage.
When ChatGPT notices that you prefer a certain tone or style during a conversation, it will proactively suggest updating these preferences without needing to manually enter the settings.
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/2006f88a-04bd-47d1-95ba-c965fe8d54e1.jpeg?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/2006f88a-04bd-47d1-95ba-c965fe8d54e1.jpeg"/>
Meanwhile, these preferences can be adjusted or removed at any time.
The updated style and tone options will be gradually rolled out starting today. The feature to fine-tune specific characteristics will be experimentally opened to some users later this week.
Both features will continue to be refined over time.
Additionally, the updated GPT-5.1 is also better at following &#34;custom instructions,&#34; giving you more precise control over tone and behavior.
Currently, updates in personalized settings will take effect immediately across all chats, including ongoing conversations, ensuring a consistent experience.
It is also important to note that changes to the base style, tone, or custom instructions only apply to newly initiated conversations.
In summary, the significant updates to GPT-5.1 and the new customization options represent an important step for OpenAI towards a ChatGPT that understands you better, is smarter, more conversational, and can adapt to personal preferences.
<h2>Available Now</h2>
Starting today, GPT-5.1 Instant and Thinking will be gradually rolled out to paid users (Pro, Plus, Go, Business), followed by free and unregistered users.
Later this week, these two models will also be available on the API. Among them, GPT-5.1 Thinking will be released under the name &#34;GPT-5.1,&#34; while GPT-5.1 Instant will be &#34;gpt-5.1-chat-latest,&#34; both equipped with adaptive reasoning capabilities.
Over the next three months, GPT-5 (Instant and Thinking) will continue to be available to paid users in the &#34;old model&#34; dropdown menu of ChatGPT, allowing users to compare and adapt at their own pace
It is worth mentioning that OpenAI indeed angered users by taking down all old models last time and has learned a painful lesson from it.
They stated that in the future, when OpenAI launches new ChatGPT models, the strategy will be to provide users with ample space to evaluate changes and share feedback, while continuously innovating cutting-edge models during a smooth transition. The phasing out of models will be clearly communicated in advance.
Author of this article: Xin Zhiyuan, Hao Kun, Tao Zi, Source: Xin Zhiyuan, Original title: &#34;GPT-5.1 Surprise Attack at Dawn, Ultraman Listens to Advice! The Human Touch Everyone is Calling for is Back&#34;
Risk Warning and Disclaimer
The market has risks, and investment requires caution. This article does not constitute personal investment advice and does not take into account the specific investment goals, financial situation, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article align with their specific circumstances. Investment based on this is at one's own risk

OpenAI

MSFT

OpenAI released GPT-5.1 this morning, bringing significant upgrades and launching three versions: GPT-5.1 Instant, GPT-5.1 Thinking, and the upcoming GPT-5.1 Pro. The new version shows significant improvements in instruction adherence and adaptive thinking, with enhancements in overall intelligence and communication style. The emotional and intellectual quotients of GPT-5.1 Instant surpass those of GPT-4, allowing users to adjust ChatGPT's tone through control options to meet different chat style needs

- OpenAI has launched GPT-5.1, a significant upgrade to the GPT-5 series, featuring three versions: Instant, Thinking, and Pro.  
- The Instant model is designed for better instruction adherence and a more conversational tone, while the Thinking model enhances reasoning capabilities and response clarity.  
- These updates aim to create a more intelligent and personable ChatGPT, allowing users to customize the model's tone and style to better suit their preferences.

Wallstreetcn

Microsoft

GPT-5.1 launched a surprise attack in the early morning, Ultraman listened to the advice! The human touch that everyone on the internet has been calling for has returned