OpenAI upgrades ChatGPT with GPT-5.5 Instant, slashing hallucinations by over half

OpenAI claims ChatGPT's new default model hallucinates way less

OpenAI's newest default model for ChatGPT might not make stuff up as much. Hallucinations have been an ongoing problem for AI models, but OpenAI says its new GPT-5.5 Instant model has "significant improvements in factuality across the board." The company claims that, based on "internal evaluations," GPT-5.5 Instant produced "52.5% fewer hallucinated claims" than its Instant model for GPT-5.3 "on high-stakes prompts covering areas like medicine, law, and finance." GPT-5.5 Instant also "reduced inaccurate claims by 37.3% on especially challenging conversations users had flagged for factual errors." OpenAI also claims that GPT-5.5 Instant is "more capable across everyday tasks," like analyzing image uploads and knowing when to turn to the web for an answer. GPT-5.5 Instant has "tighter and more to-the-point" responses and will avoid using "gratuitous emojis." With GPT-5.5 Instant, ChatGPT is now "more effective" at pulling in context from things like previous chats and your Gmail to give you more personalized responses, too. (This is a feature that Google is investing heavily in for Gemini as well.) And for all ChatGPT models, a new "memory sources" feature will let the chatbot show what context was used to inform personalized responses, and you can delete or correct information if you need. OpenAI will start rolling out GPT-5.5 Instant on Tuesday to "all ChatGPT users," though GPT-5.3 Instant will be an option for three months until it's "retired." (In the past, users have mourned the loss of older models, so this gives people time to transition.) The enhanced personalization will roll out first to Plus and Pro users on the web and is "coming soon" to the mobile apps. OpenAI has "plans" to bring it "soon" to Free, Go, Business, and Enterprise users. The memory sources feature is rolling out to ChatGPT consumer plans now on the web "and soon on mobile."

I tested GPT-5.5 Instant -- and it finally stopped overexplaining everything

I asked a simple question and for the first time in a long time, ChatGPT didn't over explain Just when I was getting used to ChatGPT-5.5 Thinking, OpenAI launches ChatGPT-5.5 Instant. Rolling out today, this model promises shorter, faster answers. More power and more speed, without the "yapping" that ChatGPT is known for. That's why this one feels noticeably different. In my testing, GPT-5.5 Instant consistently gave tighter, more focused responses. When I asked for quick feedback, it didn't default to a mini essay. When I wanted depth, it still delivered but only when it made sense. OpenAI says the model uses roughly 30% fewer words in certain scenarios, and that tracks with what I saw. The bigger difference is that it seems to understand intent better. It knows when you want a deep dive and when you just want a straight answer. If you've ever found yourself skimming past the first half of an AI response, this is the fix. A difference you'll notice The most interesting change isn't speed or brevity, but self-awareness. GPT-5.5 Instant is significantly less likely to hallucinate, especially in high-stakes areas like health, finance and legal questions. But what stood out to me wasn't just fewer mistakes. It was how the model handled them. The model actually catches itself. In one math test, I watched the model start down the wrong path, something older versions would confidently double down on. But this time, the model paused, flagged the inconsistency and corrected its own work before finishing the answer. For me, the "wait, something's off" moment, is a big deal. It's why I created the "glitch" prompt. But now, we may not need it. ChatGPT-5.5 Instant actually corrects it's course in real time, rather than confidently guessing. If you ask me, this feature is long overdue. You can now see why it knows you OpenAI is also pushing further into personalization, but with more transparency. With the new Memory Sources feature, GPT-5.5 Instant can pull from past chats, files and connected tools to tailor responses. The difference now is that you can see where that context is coming from. That means it recommends something based on your preferences, but you're not left wondering why it suggested something. Plus, you can view, edit or remove the source entirely. It's a small addition that makes the whole experience feel less like a black box. The AI assistant delivers and the performance gains are noticeable. For those keeping track of benchmarks, GPT-5.5 Instant shows strong improvements across math, science and visual reasoning tests. But honestly, that's not what stood out in day-to-day use. For me, the real upgrade I notice prompt after prompt is behavioral. It feels more controlled and less eager to impress you with volume. The takeaway After spending time with GPT-5.5 Instant, the biggest takeaway is that it is easier to use. I think you'll notice, too. You don't have to fight through extra text. You don't have to double-check every answer with the same level of skepticism. You should still check, but you don't have to over-engineer your prompts just to get something correct or concise. It feels like the model is doing more of the work behind the scenes so you don't have to. Having what feels like a "self aware" model means it knows when to stop talking and when to fix something (including its own errors.). ChatGPT-5.5 Instant is replacing ChatGPT-5.3 Instant and is rolling out to users in all tiers today. Paid users will still have GPT-5.3 Instant for three months via model configuration settings before it is retired. This is the upgrade we've been waiting for. Follow Tom's Guide on Google News and add us as a preferred source to get our up-to-date news, analysis, and reviews in your feeds. Subscribe to Tom's Guide on YouTube and follow us on TikTok.

ChatGPT gets more personal by default

Why it matters: Even subtle changes to a chatbot's tone, accuracy or memory can trigger backlash. Driving the news: OpenAI updated ChatGPT's default model to respond with more accuracy, more personalization and fewer gratuitous emoji, the company said Tuesday. * GPT-5.5 Instant is rolling out to all ChatGPT users and to the API. Enhanced personalization is initially for Plus and Pro on the web. * Free, Go, Business and Enterprise will come later. * Paid users can keep GPT-5.3 Instant for three months. By the numbers: OpenAI says GPT-5.5 Instant produced 52.5% fewer hallucinated claims than GPT-5.3 Instant on high-stakes prompts in areas like medicine, law and finance. * It also reduced inaccurate claims by 37.3% in especially challenging conversations users had flagged for factual errors. Zoom in: The new model "matches the scale of the task," the company said in a blog post Tuesday. * For casual advice prompts, the new version avoids unnecessary follow-up questions and formatting that made responses feel cluttered. Between the lines: Instant will also draw on more of a user's context in responses to prompts. * That includes information from past chats, files users have uploaded and Gmail, if connected. Reality check: OpenAI says this means users won't have to repeat themselves as often, but not everyone wants a chatbot with persistent memory. * The company is also adding "memory sources," a control that shows users some of the context ChatGPT used to personalize an answer, such as saved memories or past chats. * OpenAI says users can delete or correct outdated memory and use temporary chats that do not use or update memory. Yes, but: The increased memory serves as a reminder that prompts may be stored by AI services, depending on the service and settings. * Connecting your chatbot to third-party services means putting more of your personal or work information at risk. * The tradeoff between efficiency and privacy online has been a constant battle ever since there was an internet. Zoom out: Lower hallucination rates can also create new problems. * Users may trust answers more even when the model is still capable of getting things wrong. What we're watching: Whether OpenAI's new memory-source controls are enough to reassure users who want more personalized answers without feeling watched.

OpenAI rolls out ChatGPT 5.5 Instant as the new default model for everyone

Last week, OpenAI managed to stop ChatGPT from talking about goblins all the time. This week, there's a whole new model for users to play with. The company announced in a blog post on Tuesday that ChatGPT 5.5 Instant has begun rolling out to all users as the new default model for the popular AI chatbot. The new model is a follow-up to GPT 5.5, which was released in April. GPT-5.5 Instant replaces 5.3 Instant, which will remain available for the next three months for paid users but will otherwise be sunsetted. Unlike Claude Opus 4.7 from Anthropic and GPT-5.5, which are only available to paid customers, GPT-5.5 Instant is "available to everyone." OpenAI says it should produce fewer hallucinations and better overall results for everyday ChatGPT usage. "This update makes everyday interactions more useful and more enjoyable: stronger and tighter answers across subject areas, a more natural conversational tone, and better use of the context you've already shared when personalization can help," OpenAI's blog post said. According to OpenAI, GPT-5.5 Instant produced 52.5 percent fewer hallucinated claims in internal testing than GPT-5.3 in "high stakes" topics like law, finance, and medicine. In addition, the new model "reduced inaccurate claims by 37.3% on especially challenging conversations users had flagged for factual errors." The company also says the new model is better at deciding when to use web search for a prompt and analyzing image uploads than before. The new model is also allegedly more concise in its answers, while also maintaining something of a personality in how it talks to the user. GPT-5.5 Instant should also be better at understanding and referencing context from a connected Gmail account and other integrations to provide quality answers. And, again, most importantly, it should avoid mentioning goblins unless absolutely necessary. Want to learn more about getting the best out of your tech? Sign up for Mashable's Top Stories and Deals newsletters today.

OpenAI Just Upgraded ChatGPT's Default Model -- Here's What GPT-5.5 Instant Actually Does - Decrypt

GPT-5.5 Instant is the first Instant-tier model OpenAI classifies as "High Capability" in both cybersecurity and biological domains, requiring additional safeguards at deployment. OpenAI just swapped out the engine inside ChatGPT. Starting today, GPT-5.5 Instant replaces GPT-5.3 Instant as the default model used by hundreds of millions of people who open ChatGPT every day. This isn't a flashy launch; no new mode, no jaw-dropping demo. But "small improvement" is a relative term when the upgrade cuts hallucinations by more than half. OpenAI's GPT family ships in tiers. Instant is the everyday model, built for speed and general use; Thinking is the slower, more analytical version for complex problems; and Pro is the heavyweight for maximum-intensity tasks. GPT-5.5 Instant is the latest update to the tier that most ChatGPT users will interact with, whether they realize it or not. According to OpenAI, the new model produced fewer hallucinated claims than GPT-5.3 Instant on high-stakes prompts in medicine, law, and finance. Hallucinations have been ChatGPT's most persistent flaw since the beginning. OpenAI also tested against conversations real users had previously flagged for factual errors. On those, inaccurate claims dropped by 37.3%. On HealthBench -- a benchmark testing AI responses to real medical questions, scored from 0 to 100 -- GPT-5.5 Instant scores 51.4 points, up from 49.6. On HealthBench Professional, the clinical-use version, it jumps from 32.9 to 38.4 points. Health questions are among the most common things people ask ChatGPT, which makes getting them right more than a benchmark exercise. These results mean GPT 5.5 Instant increased accuracy by responding correctly 38.4% of the time. GPT-5.5 Instant also pulls more actively from your past chats, saved files, and connected Gmail account to make answers personally relevant. Now when it does this, it shows you exactly what context it used, and lets you delete or correct it. "You remain in control of what's in your memory," OpenAI wrote. Temporary chats still opt out entirely. When Decrypt covered the GPT-5.5 family launch two weeks ago, the story was agentic coding and terminal workflows. GPT-5.5 Instant is a different animal -- it handles more "basic" stuff like your meal plans and email drafts, not autonomous multi-step coding pipelines. Don't ask us about GPT 5.4 Instant, though. It's probably chilling next to the O2 model that never existed. The full GPT-5.5 scores 82.7% on Terminal-Bench 2.0, which measures complex command-line task performance. Instant is what the rest of us get, and probably what most of the users will probably be fine working with. One notable footnote in the system card: GPT-5.5 Instant is the first Instant-tier model OpenAI classifies as "High Capability" in both cybersecurity and biological domains -- capable enough to require the same automated safeguards previously reserved for the more powerful Thinking variants. It won't help you hack anything, but OpenAI built guardrails in case someone tries. The previous default, GPT-5.3 Instant, launched in March with promises of fewer preachy refusals and better accuracy. GPT-5.5 Instant continues that trajectory. Paid subscribers who prefer the old version have three months before GPT-5.3 Instant is retired. Enhanced personalization via Gmail rolls out first to Plus and Pro users on the web, with Free, Go, Business, and Enterprise to follow in the coming weeks.