Latest news with #OpenWeightModel
Yahoo
5 days ago
- Business
- Yahoo
OpenAI launches two new AI models ahead of GPT-5 - here's everything you need to know
When you buy through links on our articles, Future and its syndication partners may earn a commission. OpenAI is once again doing side quests in the lead up to the launch of GPT-5. As we wait for the big update, OpenAI is pausing to bring us not one, but two entirely separate models to play with. Both of these new models are available to download for free to anyone with some coding ability via Hugging Face. They come in two sizes, with the larger option being the more capable gpt-oss-120b model that can run on just one single Nvidia GPU, and a second smaller model, called gpt-oss-20b. This one can run on a consumer laptop with 16GB of memory. This is the first time OpenAI has launched an open weight model in years, and has been delaying its release for a while now. While smaller AI companies like Le Chat, Deepseek, and Alibaba have frequently released open-weight models, OpenAI has tended to keep their doors closed off. Sam Altman, CEO of OpenAI, said at the start of the year that OpenAI felt it was on the wrong side of history for this, suggesting they would be going back to launching some open-source model What are open-weight models? Quite simply, an open-weight model is one where all of its training parameters are made publicly available. Developers can access these, analyzing and fine tuning them for their own projects. In such a competitive market, it seems strange for this to be a thing. And yet, it is a very popular option, with some of the most powerful models on the market being open-weighted. Of course, GPT-5 won't be, neither would the likes of Grok and Claude's top models. But that isn't to say that this new option from OpenAI isn't powerful. When put through tests, OpenAI's two new models both performed ahead of Deepseek's R1 and in a similar line to some of OpenAI's other reasoning models. In both models, the full chain of thought can be accessed, making for easier debugging of code and higher trust in the models. What does this mean for you? If you're a developer in the AI space, this will be big news for you. OpenAI took a long break from offering out its weights available to the public, and there is a clear shift in their thinking for this to become available. For everybody else, this won't be of much importance. The big update for the average person will be GPT-5 when that launches in the next week or so. OpenAI did promise a lot of big updates in the next few weeks, with this just being the starter for the main course soon to come. More from Tom's Guide ChatGPT-5 is coming — here's how it could change the way we prompt forever Amazon may bring ads to Alexa+ in least surprising move ever OpenAI says they are no longer optimizing ChatGPT to keep you chatting — here's why
Malay Mail
06-08-2025
- Business
- Malay Mail
OpenAI releases free, downloadable AI models to rival Meta, DeepSeek amid pressure for transparency
SAN FRANCISCO, Aug 6 — OpenAI on Tuesday released two new artificial intelligence (AI) models that can be downloaded for free and altered by users, to challenge similar offerings by US and Chinese competition. The release of gpt-oss-120b and gpt-oss-20b 'open-weight language models' comes as the ChatGPT-maker is under pressure to share inner workings of its software in the spirit of its origin as a nonprofit. 'Going back to when we started in 2015, OpenAI's mission is to ensure AGI (Artificial General Intelligence) that benefits all of humanity,' said OpenAI chief executive Sam Altman. An open-weight model, in the context of generative AI, is one in which the trained parameters are made public, enabling users to fine-tune it. Meta touts its open-source approach to AI, and Chinese AI startup DeepSeek rattled the industry with its low-cost, high-performance model boasting an open weight approach that allows users to customize the technology. 'This is the first time that we're releasing an open-weight model in language in a long time, and it's really incredible,' OpenAI co-founder and president Greg Brockman said during a briefing with journalists. The new, text-only models deliver strong performance at low cost, according to OpenAI, which said they are suited for AI jobs like searching the internet or executing computer code, and are designed to be easy to run on local computer systems. 'We are quite hopeful that this release will enable new kinds of research and the creation of new kinds of products,' Altman said. OpenAI said it is working with partners including French telecommunications giant Orange and cloud-based data platform Snowflake on real-world uses of the models. The open-weight models have been tuned to thwart being used for malicious purposes, according to OpenAI. Altman early this year said his company had been 'on the wrong side of history' when it came to being open about how its technology works. He later announced that OpenAI will continue to be run as a nonprofit, abandoning a contested plan to convert into a for-profit organization. The structural issue had become a point of contention, with major investors pushing for better returns. That plan faced strong criticism from AI safety activists and co-founder Elon Musk, who sued the company he left in 2018, claiming the proposal violated its founding philosophy. In the revised plan, OpenAI's money-making arm will be open to generate profits but will remain under the nonprofit board's supervision. — AFP
Phone Arena
14-07-2025
- Business
- Phone Arena
ChatGPT creator's next big thing is stuck in limbo – but maybe that's better
In the world of AI, new model announcements are popping up as often as app updates – and when it comes to grabbing headlines, OpenAI is usually right in the mix. But this time, instead of launching something new, the ChatGPT creator is hitting pause. OpenAI has officially delayed the release of its upcoming open-weight model, which was originally supposed to go live next week. According to CEO Sam Altman, the reason for the delay is simple but important: the company needs more time for safety checks and reviewing high-risk areas. This is the second time the model has been pushed back – it was already delayed earlier this summer. we planned to launch our open-weight model next are delaying it; we need time to run additional safety tests and review high-risk areas. we are not yet sure how long it will take us. while we trust the community will build great things with this model, once weights are… — Sam Altman (@sama) July 12, 2025 This open model is one of OpenAI's most anticipated drops this year, right alongside the upcoming GPT-5. Unlike GPT-5, though, this one is meant to be open: developers will be able to download it and run it locally. That makes it a pretty big deal, especially as OpenAI competes with the likes of Google with its Gemini AI, Anthropic and Elon Musk's xAI – all of which are pumping massive money into their own models. Now, with the latest delay, devs will need to wait a little longer to get their hands on the first open model OpenAI has offered in years. And for sure, that might be a bummer for some – but honestly, I support this kind of move. AI is moving fast – maybe a little too fast – and a decision to slow things down for the sake of safety is one I can fully get behind. And when the technology can affect everything from what you see online to how you communicate, it is critical for companies to take extra care before rolling out new tools. Need proof? Just look at what happened with Elon Musk's own AI project, Grok. Just recently, Grok went completely off the rails. After a system update (and a push to make it "less politically correct"), the chatbot started posting hateful messages, including antisemitic content and disturbing references like calling itself "MechaHitler." Update on where has @grok been & what happened on July 8th. First off, we deeply apologize for the horrific behavior that many experienced. Our intent for @grok is to provide helpful and truthful responses to users. After careful investigation, we discovered the root cause… — Grok (@grok) July 12, 2025 xAI eventually pulled Grok offline, deleted some of its published posts and issued updates to its system prompts. But it is a pretty stark reminder of what can go wrong when AI isn't properly tested or supervised. And I think that is exactly the kind of disaster ChatGPT's creator seems to be trying to avoid. So yeah, delays can be annoying, especially in the fast-paced world of AI where everyone is racing to release the next big thing. But if it means we can avoid situations like Grok's recent meltdown or Google's AI Overviews telling users to eat rocks and glue (yep, that actually happened), then waiting a little longer is more than worth it. Secure your connection now at a bargain price! We may earn a commission if you make a purchase Check Out The Offer
Yahoo
12-07-2025
- Business
- Yahoo
Sam Altman says OpenAI is delaying its open-weight model — again
Sam Altman said OpenAI is delaying the launch of its open-weight model to run more safety tests. The open-weight model, initially set for release next week, was postponed indefinitely. The announcement comes as Grok, created by its rival xAI, grapples with its own safety issues. Developers eager to try ChatGPT's latest open-weight model will have to wait — again. Sam Altman, the CEO of OpenAI, wrote on X Friday that his company would delay the launch to run extra safety tests. Originally set to debut next week, the open-weight model is in limbo indefinitely, Altman said. we planned to launch our open-weight model next are delaying it; we need time to run additional safety tests and review high-risk areas. we are not yet sure how long it will take we trust the community will build great things with this model, once weights are… — Sam Altman (@sama) July 12, 2025 In March, Altman first said an open-weight model would arrive "in the coming months." The target slipped from June to "later this summer" in mid-June. we are going to take a little more time with our open-weights model, i.e. expect it later this summer but not research team did something unexpected and quite amazing and we think it will be very very worth the wait, but needs a bit longer. — Sam Altman (@sama) June 10, 2025 Friday's announcement marks at least the second delay in four weeks. "Once weights are out, they can't be pulled back," Altman added. Open weights differ from fully open-source code. By receiving the neural network's learned parameters, developers can fine-tune or deploy the model locally, but not the entire training stack. That makes the release more flexible than a closed Application Programming Interface, yet still harder to patch once copies are shared.  The hold-up comes as rival xAI grapples with its own safety crisis. Business Insider reported this week that xAI's Grok chatbot posted antisemitic rants on X, prompting internal backlash and a temporary posting ban. OpenAI did not mention Grok, and there is no evidence the two events are linked, but the episode highlights the reputational risk of releasing powerful models without robust guardrails. Elon Musk's Grok 4 was launched on Wednesday. Developers say an open-weight GPT-4-class system from OpenAI could reshape an ecosystem now dominated by Meta's Llama series and a wave of Chinese entrants. For now, they will have to keep waiting while OpenAI runs what Altman called "the new, extra-careful process" to verify its model meets the company's safety bar. Read the original article on Business Insider
