OpenAI's New GPT 4.1 Models Excel at Coding
Apr 14, 2025 1:40 PM GPT 4.1, GPT 4.1 Mini, and GPT 4.1 Nano are all available now—and will help OpenAI compete with Google and Anthropic. PHOTO COLLAGE: J.D. REEVES; GETTY IMAGES
OpenAI announced today that it is releasing a new family of artificial intelligence models optimized to excel at coding, as it ramps up efforts to fend off increasingly stiff competition from companies like Google and Anthropic. The models are available to developers through OpenAI's application programming interface (API).
OpenAI is releasing three sizes of models: GPT 4.1, GPT 4.1 Mini, and GPT 4.1 Nano. Kevin Weil, chief product officer at OpenAI, said on a livestream that the new models are better than OpenAI's most widely used model, GPT-4o, and better than its largest and most powerful model, GPT-4.5, in some ways.
GPT-4.1 scored 55 percent on SWE-Bench, a widely used benchmark for gauging the prowess of coding models. The score is several percentage points above that of other OpenAI models. The new models are 'great at coding, they're great at complex instruction following, they're fantastic for building agents,' Weil said.
The capacity for AI models to write and edit code has improved significantly in recent months, enabling more automated ways of prototyping software, and improving the abilities of so-called AI agents. In the past few months, rivals like Anthropic and Google have both introduced models that are especially good at writing code.
The arrival of GPT-4.1 has been widely rumored in recent weeks. OpenAI apparently tested the model on some popular leaderboards under the pseudonym Alpha Quasar, sources say. Some users of the 'stealth' model reported impressive coding abilities. 'Quasar fixed all the open issues I had with other code genarated [sic] via llms's which was incomplete,' one person wrote on Reddit.
'Developers care a lot about coding and we've been improving our model's ability to write functional code,' Michelle Pokrass, who works on post-training at OpenAI, said during the Monday livestream. 'We've been working on making it follow different formats and better explore repos, run unit tests and write code that compiles.'
Over the past couple of years, OpenAI has parlayed feverish interest in ChatGPT, a remarkable chatbot first unveiled in late 2022, into a growing business selling access to more advanced chatbots and AI models. In a TED interview last week, Altman said that OpenAI had 500 million weekly active users, and that usage was 'growing very rapidly.'
OpenAI now offers a smorgasbord of different flavors of models with different capabilities and different pricing. The company's largest and most powerful model, called GPT-4.5, was launched in February, though OpenAI called the launch a 'research preview' because the product is still experimental.
The company also offers models called o1 and o3 that are capable of performing a simulated kind of reasoning, breaking a problem down into parts in order to solve it. These models also take longer to respond to queries and are more expensive for users.
ChatGPT's success has inspired an army of imitators, and rival AI players have ramped up their investments in research in an effort to catch up to OpenAI in recent years. A report on the state of AI published by Stanford University this month found that models from Google and DeepSeek now have similar capabilities to models from OpenAI. It also showed a gaggle of other firms including Anthropic, Meta, and the French firm Mistral in close pursuit.
Oren Etzioni, a professor emeritus at the University of Washington who previously led the Allen Institute for AI (AI2), says it is unlikely that any single model or company will be dominant in the future. 'We will see even more models over time as cost drops, open source increases, and specialized models win out in different arenas including biology, chip design, and more,' he says. Etzioni adds that he would like to see companies focus on reducing the cost and environmental impact of training powerful models in the years ahead.
OpenAI faces pressure to show that it can build a sustained and profitable business by selling access to its AI models to other companies. The company's chief operating officer, Brad Lightcap, told CNBC in February that the company had more than 400 million weekly active users, a 30 percent increase from December 2024. But the company is still losing billions as it invests heavily in research and infrastructure. In January, OpenAI announced that it would create a new company called Stargate in collaboration with SoftBank, Oracle, and MGX. The group collectively promised to invest $500 billion in new AI datacenter infrastructure.
In recent weeks, OpenAI has teased a flurry of new models and features. Last week, Altman announced that ChatGPT would receive a memory upgrade allowing the chatbot to better remember and refer back to previous conversations. In late March, Altman announced that OpenAI plans to release an open weight model, which developers will be able to download and modify for free, in the summer. The company said it would begin testing the model in the coming weeks. Open weight models are already popular with researchers, developers, and startups because they can be tailored for different uses and are often cheaper to use.
This is a developing story. Please check back for updates .
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Tom's Guide
an hour ago
- Tom's Guide
This ChatGPT ‘memory hack' changes everything — use these prompts to make it remember you
If you've ever found yourself reintroducing yourself, your tone preferences or even something as basic as your name to ChatGPT, you're not alone. While OpenAI's chatbot is famous for being smart, helpful and shockingly conversational, it's not always great at remembering what matters to you, which is why you might want to teach it. The good news is that ChatGPT actually has a memory feature that's smarter than most people realize, and there are a few tricks you can use to make it remember exactly what you want. Here's how to unlock ChatGPT's memory, plus a few sneaky hacks to get it acting like a real personal assistant. ChatGPT's memory is designed to remember helpful facts about you including your name, your job, your writing style and even your goals. Knowing these things can help the AI tailor responses over time. Think of it like your AI assistant building a mental file cabinet with your preferences inside. OpenAI first rolled out memory to GPT-4o users in early 2024, and now it's automatically turned on for most ChatGPT Plus users. But unless you're actively using it, or customizing it, you might not get the full benefit. To check if memory is on, go to Settings → Personalization → Memory. From there, you can view, edit or wipe everything ChatGPT has remembered about you. One of the simplest ways to store a fact in ChatGPT's memory is to literally prompt it to remember something. For example: Get instant access to breaking news, the hottest reviews, great deals and helpful tips. If memory is enabled, ChatGPT will usually respond with: 'Got it. I'll remember that for future chats.' If not, it may ask for permission to store that information. I've noticed that when I've asked ChatGPT to remember things, it doesn't always remember the first time. Sometimes, not even the second time. If you run into this problem, stay persistent and keep reminding ChatGPT to remember something until it actually does example, I once used ChatGPT Vision to help my mom match fabric for a project. From then on, ChatGPT thought I was a quilter. I had to tell the chatbot to forget that (much to my mom's dismay, I'm sure).Pro tip: You can also say 'Forget that' or 'Update my memory' if something changes — like your job or preferred tone. If you want to be sure it does not remember something, you can also use the temporary chat feature. Even if memory is off or you aren't a Plus subscriber, you can still simulate long-term memory using what I call the context chaining trick. Here's how: Start your prompt with: 'For the rest of this conversation, assume I'm a second grade teacher working on an end-of-year project for my students.' This doesn't persist across sessions, but it works surprisingly well for one-off tasks or multi-step projects. OpenAI makes it easy to see what ChatGPT has remembered — and yes, you should check it occasionally. Just type: 'What do you remember about me?' It'll respond with a summary of the info it has on file, like: 'You're a mom of three who juggles work, parenting and writing with a good sense of humor. You're no stranger to trampoline parks, fourth grade homework chaos or PTA drama. You're based in New Jersey, drive a Jeep and sometimes test AI tools for personal life (like IEP meetings or canceling gym memberships).' Here are a few fast rules to get the most out of ChatGPT's memory: ChatGPT won't magically know your preferences unless you teach it, but with memory, it can get surprisingly close. Whether you're writing a novel, planning a trip or just tired of repeating yourself, these memory hacks can turn ChatGPT into a genuinely helpful sidekick.
TechCrunch
an hour ago
- TechCrunch
For the love of God, stop calling your AI a co-worker
Generative AI comes in many forms. Increasingly, though, it's marketed the same way: with human names and personas that make it feel less like code and more like a co-worker. A growing number of startups are anthropomorphizing AI to build trust fast — and soften its threat to human jobs. It's dehumanizing, and it's accelerating. I get why this framing took off. In today's upside-down economy, where every hire feels like a risk, enterprise startups — many emerging from the famed accelerator Y Combinator — are pitching AI not as software but as staff. They're selling replacements. AI assistants. AI coders. AI employees. The language is deliberately designed to appeal to overwhelmed hiring managers. Some don't even bother with subtlety. Atlog, for instance, recently introduced an 'AI employee for furniture stores' that handles everything from payments to marketing. One good manager, it gloats, can now run 20 stores at once. The implication: you don't need to hire more people — just let the system scale for you. (What happens to the 19 managers it replaces is left unsaid.) Consumer-facing startups are leaning into similar tactics. Anthropic named its platform 'Claude' because it's a warm, trustworthy-sounding companion for a faceless, disembodied neural net. It's a tactic straight out of the fintech playbook where apps like Dave, Albert, and Charlie masked their transactional motives with approachable names. When handling money, it feels better to trust a 'friend.' The same logic has crept into AI. Would you rather share sensitive data with a machine learning model or your bestie Claude, who remembers you, greets you warmly, and almost never threatens you? (To OpenAI's credit, it still tells you you're chatting with a 'generative pre-trained transformer.') But we're reaching a tipping point. I'm genuinely excited about generative AI. Still, every new 'AI employee' has begun to feel more dehumanizing. Every new 'Devin' makes me wonder when the actual Devins of the world will push back on being abstracted into job-displacing bots. Generative AI is no longer just a curiosity. Its reach is expanding, even if the impacts remain unclear. In mid-May, 1.9 million unemployed Americans were receiving continued jobless benefits — the highest since 2021. Many of those were laid-off tech workers. The signals are piling up. Techcrunch event Save now through June 4 for TechCrunch Sessions: AI Save $300 on your ticket to TC Sessions: AI—and get 50% off a second. Hear from leaders at OpenAI, Anthropic, Khosla Ventures, and more during a full day of expert insights, hands-on workshops, and high-impact networking. These low-rate deals disappear when the doors open on June 5. Exhibit at TechCrunch Sessions: AI Secure your spot at TC Sessions: AI and show 1,200+ decision-makers what you've built — without the big spend. Available through May 9 or while tables last. Berkeley, CA | REGISTER NOW Some of us still remember 2001: A Space Odyssey. HAL, the onboard computer, begins as a calm, helpful assistant before turning completely homicidal and cutting off the crew's life support. It's science fiction, but it hit a nerve for a reason. Last week, Anthropic CEO Dario Amodei predicted that AI could eliminate half of entry-level white-collar jobs in the next one to five years, pushing unemployment as high as 20%. 'Most [of these workers are] unaware that this is about to happen,' he told Axios. 'It sounds crazy, and people just don't believe it.' You could argue that's not comparable to cutting off someone's oxygen, but the metaphor isn't that far off. Automating more people out of paychecks will have consequences, and when the layoffs increase, the branding of AI as a 'colleague' is going to look less clever and more callous. The shift toward generative AI is happening regardless of how it's packaged. But companies have a choice in how they describe these tools. IBM never called its mainframes 'digital co-workers.' PCs weren't 'software assistants'; they were workstations and productivity tools. Language still matters. Tools should empower. But more and more companies are marketing something else entirely, and that feels like a mistake. We don't need more AI 'employees.' We need software that extends the potential of actual humans, making them more productive, creative, and competitive. So please stop talking about fake workers. Just show us the tools that help great managers run complex businesses. That's all anyone is really asking for.
Business Insider
2 hours ago
- Business Insider
Here's how Uber's product chief uses AI at work — and one tool he's going to use next
Uber's chief product officer has one AI tool on his to-do list. In an episode of "Lenny's Podcast" released on Sunday, Uber's product chief, Sachin Kansal, shared two ways he is using AI for his everyday tasks at the ride-hailing giant and how he plans to add NotebookLM to his AI suite. Kansal joined Uber eight years ago as its director of product management after working at cybersecurity and taxi startups. He became Uber's product chief last year. Kansal said he uses OpenAI's ChatGPT and Google's Gemini to summarize long reports. "Some of these reports, they're 50 to 100 pages long," he said. "I will never have the time to read them." He said he uses the chatbots to acquaint himself with what's happening and how riders are feeling in Uber's various markets, such as South Africa, Brazil, and Korea. The CPO said his second use case is treating AI like a research assistant, because some large language models now offer a deep research feature. Kansal gave a recent example of when his team was thinking about a new driver feature. He asked ChatGPT's deep research mode about what drivers may think of the add-on. "It's an amazing research assistant and it's absolutely a starting point for a brainstorm with my team with some really, really good ideas," the CPO said. In April, Uber's CEO, Dara Khosrowshahi, said that not enough of his 30,000-odd employees are using AI. He said learning to work with AI agents to code is "going to be an absolute necessity at Uber within a year." Uber did not immediately respond to a request for comment from Business Insider. Kansal's next tool: NotebookLM On the podcast, Kansal also highlighted NotebookLM, Google Lab's research and note-taking tool, which is especially helpful for interacting with documents. He said he doesn't use the product yet, but wants to. "I know a lot of people who have started using it, and that is the next thing that I'm going to use," he said. "Just to be able to build an audio podcast based on a bunch of information that you can consume. I think that's awesome," he added. Kansal was referring to the "Audio Overview" feature, which summarizes uploaded content in the form of two AIs having a voice discussion. NotebookLM was launched in mid-2023 and has quickly become a must-have tool for researchers and AI enthusiasts. Andrej Karpathy, Tesla's former director of AI and OpenAI cofounder, is among those who have praised the tool and its podcast feature. "It's possible that NotebookLM podcast episode generation is touching on a whole new territory of highly compelling LLM product formats," he said in a September post on X. "Feels reminiscent of ChatGPT. Maybe I'm overreacting."
