Follow ZDNET: Add us as a favorite source On Google.
ZDNET Highlights
- GPT-5.1 speeds up coding with adaptive and logic-free modes.
- New prompt caching cuts API costs for embedded app developers.
- New tools make AI agents more capable inside modern IDEs.
OpenAI is back with a new 5.1 update to its previous GPT-5 large language model. GPT-5 was introduced in August, a time warp-speed version of our universe’s AI from decades ago.
Of course, OpenAI is using AI to help code faster. After all, it is in a race with other big players to achieve that trillion-dollar valuation. Furthermore, it has been proven without any doubt that AI coding, in the hands of a professional coder, is an almost magical force multiplier and project accelerator.
(Disclosure: ZDNET’s parent company Ziff Davis filed a lawsuit against OpenAI in April 2025, alleging it infringed Ziff Davis copyrights in the training and operation of its AI systems.)
Also: OpenAI’s GPT-5.1 makes ChatGPT ‘warmer’ and smarter – how its advanced modes now work
For an overview of the benefits of GPT-5.1 for consumer chatbot users, read Senior Editor Sabrina Ortiz’s explainer. But if you’re interested in using AI in your coding or embedded in your software, keep reading. There are some solid speed and cost-saving benefits in this release.
In this article we are talking about GPT-5.1 in APIIn other words, we are looking at sending signals to the AI through the program’s function call and receiving the result as the return value of that call,
This API-driven AI functionality works within software products created by developers, but since the developer tools themselves also use APIs to provide intelligence, it extends the usefulness of those tools. This also benefits developers using OpenAI’s Codex coding agent, as the Codex is now available in release 5.1.
Also: The Best Free AI Courses and Certifications for Upskilling in 2025 – and I’ve Tried Them All
For example, JetBrains is a manufacturer of excellent development tools. Although I moved away from the JetBrains platform because VS Code is so widely used (and I often need to talk to you about it), JetBrains products are still some of my favorites. In fact, using VS Code, I sometimes miss some of JetBrains’ features.
That’s why it was so interesting when Denis Shiryaev, head of the AI DevTools Ecosystem at JetBrains, described the company’s experience with this new GPT-5.1 release in an OpenAI blog post. He said, “GPT 5.1 is not just another LLM – it is truly agentic, the most inherently autonomous model I have ever tested.”
“It writes like you, codes like you, follows complex instructions intuitively, and excels at front-end tasks, all while fitting nicely into your existing codebase,” he said.
Let’s take a look at some of the reasons why GPT-5.1 is receiving such an enthusiastic response.
adaptive logic
I found coding with GPT-5 surprisingly powerful, but sometimes tedious. No matter what I asked the AI, the response took time. Even the simplest question may take a few minutes to get answered. This is because all queries sent requests to the same model.
GPT-5.1 evaluates the given prompt and, depending on whether the question is inherently easy or difficult, adjusts how much cognitive effort it puts into the answer. This means that simple queries will no longer experience the delays that were so frustrating when using older coding models.
Here’s a prompt I gave to GPT-5 a few days ago: “Please check my work. I’m renaming EDD_SL_Plugin_Updater so that each plugin that uses it has a unique name to avoid conflicts. I updated the class name in the updater file, updated the updater file name, and then updated the file and class references in the plugin’s main file. Can you check the plugins and make sure there are no errors? If you see anything, If found, report back to me and make no changes.”
Also: 10 ChatGPT Prompt Tricks I Use – To Get Best Results Fast
This is a huge request, which requires the AI to scan 12,000 files and give me an analysis. It should use all its thinking power.
In contrast, “Does the WP-CLI command show a list of installed plugins?” Simple request really. It’s basically a documentation lookup that doesn’t require any real intelligence. This is just a quick time-saving hint so I don’t have to switch browsers and do a Google search.
Answers to quick queries are faster, and the process uses fewer tokens. Token is a measure of the amount of processing used. API calls are charged on a token basis, meaning it will cost less to make simple feature queries.
There’s another aspect to this that’s quite powerful, which OpenAI describes as “more continuous deep reasoning.” There’s nothing more wasteful than having a long conversation with an AI and then losing track of what you were talking about. Now, OpenAI says AI can stay on track for a long time.
‘No arguments’ mode
This is one of those cases where I think OpenAI could benefit from some solid product management for its product naming. This mode does not turn off context understanding, quality code writing, or understanding instructions. It simply shuts down deep, chain-of-thought-style analysis. They should call it “don’t think too much” mode.
Think of it this way. We all have that friend who overthinks about every issue or task. This makes them nervous, takes them a lot of time to do simple tasks and often leads to analysis paralysis. It’s time to think big, and also a time to pick up the paper or plastic and move on.
Also: I linked two AI tools together to solve a major bug – but they couldn’t do it without me
This new logic-free mode enables the AI to avoid its usual step-by-step deliberations and simply arrive at an answer. It is ideal for simple lookups or basic tasks. This dramatically cuts latency (time to respond). It also creates a more responsive, quicker, and more fluid coding experience.
Combining unbiased reasoning with adaptive reasoning means AI can take time to answer difficult questions, but can answer simple questions faster.
extended prompt caching
Another increase in speed (along with a reduction in cost) is Extended Accelerated Caching. When AI is given a signal, it first has to use its natural language processing capabilities to parse that signal to figure out what it is being asked.
This is no small achievement. It took decades for AI researchers to get AI to a level where they could understand natural language, as well as the context and subtle meanings of what was being said.
So, when a signal is issued, the AI has to do some real work to tokenize it, to create an internal representation from which to formulate a response. This is not without its resource usage costs.
Also: 10 ChatGPT codec secrets I learned after just 60 hours of pair programming with it
If a question is asked again during a session, and the same or similar prompt has to be interpreted again, that cost is incurred again. Keep in mind that we’re talking not only about signals that a programmer returns to an API, but also about signals that run inside an application, which may often be repeated during use of the application.
For example, take a detailed prompt for a customer support agent, who has to process the same set of basic initiation rules for every customer interaction. Parsing that signal could take thousands of tokens, and would need to be done thousands of times a day.
By caching the prompt (and OpenAI is now doing this for 24 hours), the prompt is compiled once and then available for reuse. The speed improvements and cost savings can be substantial.
A better business case for design-ins
All of these improvements provide a better business case for OpenAI to present to customers for design-in. Design-in is a fairly old term of art, used to describe when a component is designed into a product.
Probably the most famous (and most consequential) design was when IBM chose the Intel 8088 CPU for the original IBM PC in 1981. That one decision launched the entire x86 ecosystem and fueled Intel’s success in processors for decades.
Today, Nvidia is the beneficiary of huge design-in decisions by data center operators who are most hungry for AI processing power. That demand has propelled Nvidia to become the world’s most valuable company in terms of market cap, just north of $5 trillion.
Also: I completed 4 years of product development in 4 days for $200, and I’m still shocked
OpenAI also benefits from design-ins. CapCut is a video app 361 million downloads Temu is a shopping app with 438 million downloads in 2025. For example, if any company embeds AI into their app, and if they do so using API calls from OpenAI, then OpenAI stands to earn a ton of cash from the cumulative volume of API calls and their associated billing.
But like physical components, cost of goods sold is always an issue with design-in. Every fraction of a cent in COGS can increase the overall final price or have a dangerous impact on margins.
So, the bottom line is that if OpenAI can reduce the cost of API calls significantly and still provide AI value, as it has done with GPT-5.1, then there is a better chance that it can make a case for developers to incorporate GPT-5.1 into their products.
More new capabilities
The GPT-5.1 release also includes improved coding performance. AI is more operable and biddable, which means it follows instructions better. If only my puppy could be more biddable, we wouldn’t have to constantly have painful yelps when the mail is delivered.
Coding AI does less unnecessary overthinking, is more conversational during tool-calling sequences, and behaves more overall friendly during sequence interactions. There’s also a new apply_patch tool that helps with multi-step coding sequences and agentive actions, as well as a new shell tool that does a better job when asked to generate command-line commands and evaluate and act based on responses.
Also: OpenAI now has a new agentive coding partner for you: GPT-5-Codex
I am very excited about this new release. Since I’m already using GPT-5, it would be nice to see how much more responsive it is now with GPT-5.1.
How are you? Have you tried using GPT-5 or the new GPT-5.1 model in your coding or development workflow? Are you seeing what kind of speed or cost improvements OpenAI is promising, or are you still evaluating whether these changes make sense for your projects? How important are features like adaptive logic, no logic mode, or instant caching when you’re deciding which AI models to build into your tools or products? Let us know in the comments below.
You can follow my daily project updates on social media. Be sure to subscribe My weekly update newsletterAnd follow me on Twitter/X @davidgewirtzon facebook Facebook.com/DavidGewirtzon instagram Instagram.com/DavidGewirtzon bluesky @DavidGewirtz.comand on youtube YouTube.com/DavidGewirtzTV,
