News from the AI & ML world
DeeperML - #gpt5
|
@www.marktechpost.com
//
OpenAI has launched the Evals API, a new tool designed to streamline the evaluation of large language models (LLMs) for developers and teams. The Evals API introduces programmatic evaluation capabilities, allowing developers to define tests, automate evaluation runs, and iterate on prompts directly from their workflows. Previously, evaluations were accessible only through the OpenAI dashboard, but the new API enables a more integrated and systematic approach to assessing model performance.
The Evals API aims to address the challenges of manually evaluating LLM performance, which can be time-consuming, especially when scaling applications across diverse domains. By providing a systematic approach, OpenAI hopes to improve custom test case assessments, measure improvements across prompt iterations, and automate quality assurance in development pipelines. This will enable developers to treat evaluation as a core part of their development cycle, similar to unit tests in traditional software engineering. Despite these advancements, OpenAI has announced a delay in the launch of GPT-5 by a few months. According to CEO Sam Altman, the delay is due to the company's efforts to significantly improve the model and ensure enough capacity to support expected high demand. In the meantime, OpenAI plans to release o3 and o4-mini models. The company has also faced capacity issues with current features, as seen with the restrictions placed on the image generation software after its launch. Recommended read:
References :
@Latest from Tom's Guide
//
OpenAI has announced a shift in its release strategy for GPT-5, delaying the launch by several months. CEO Sam Altman revealed the change in plans, stating that the company will first release its reasoning models, o3 and o4-mini, in the coming weeks. This reverses earlier plans to integrate these models directly into GPT-5. The full GPT-5 model is now expected to arrive "in a few months."
Altman cited several reasons for the delay. Integrating all components into a single unified system proved more challenging than initially anticipated. Additionally, the extra development time has revealed the potential to make GPT-5 "much better than we originally thought." Ensuring sufficient computing capacity to meet the expected "unprecedented demand" was also a key factor. The o3 model, in particular, has undergone significant improvements since its internal preview, with Altman stating that "people will be happy" with the advancements. The o3 and o4-mini models are classified as reasoning models, designed to perform complex thinking tasks. These models have demonstrated stronger performance than conventional language models in areas like coding and mathematics. OpenAI first introduced its o3 model in December 2024, marking a major advancement in complex reasoning tasks, followed by the more affordable and faster o3-mini version in late January 2025. While users will need to wait longer for GPT-5, the upcoming release of o3 and o4-mini promises exciting advancements in AI capabilities. Recommended read:
References :
Chris McKay@Maginative
//
OpenAI is shaking up its AI model release strategy, announcing plans to launch o3 and o4-mini in the coming weeks before the much-anticipated GPT-5. This marks a reversal from earlier plans to consolidate efforts around GPT-5. CEO Sam Altman cited technical integration challenges and the need for sufficient capacity to handle expected demand as factors influencing the decision. Altman expressed confidence that the delay will allow OpenAI to make GPT-5 "much better than we originally thought," promising substantial improvements to the flagship model.
The unexpected addition of o4-mini indicates that OpenAI isn't slowing down its pace of innovation. The release of o3 and o4-mini comes as OpenAI faces increasing competition in the AI market. In a strategic move targeting the education sector, OpenAI is now offering free ChatGPT Plus subscriptions to college students. This initiative aims to escalate competition with Anthropic, particularly following Anthropic's unveiling of "Claude for Education" and partnerships with several universities. In addition to model development, OpenAI is reportedly finalizing a significant funding deal, potentially worth $40 billion, with SoftBank. The funds are intended to further advance the capabilities of the models and address any safety concerns. The influx of capital could solidify OpenAI's position as a leading force in the rapidly evolving AI landscape, enabling them to pursue ambitious research and development projects while navigating the competitive pressures from rivals like Google and Anthropic. Recommended read:
References :
Matthias Bastian@THE DECODER
//
OpenAI is making adjustments to its AI model release strategy, with a shift concerning its highly anticipated GPT-5. Originally planned to integrate new reasoning models o3 and o4-mini, OpenAI will now release these as standalone systems in the coming weeks. This decision results in delaying the GPT-5 release by a few months.
CEO Sam Altman cited the difficulty of integrating components into a unified system as a primary factor, along with the potential for GPT-5 to exceed initial expectations. Ensuring adequate computing capacity to meet anticipated demand also played a role. Altman highlighted significant improvements in the o3 model since its initial preview. OpenAI is also making moves to increase accessibility. It is now offering free ChatGPT Plus subscriptions to college students. This aims to provide access to advanced AI tools like GPT-4o, image generation, and voice interaction. This offering coincides with Anthropic's recent introduction of "Claude for Education," setting the stage for a fierce competition in the education AI market as the tech giants battle for dominance in the $80 billion education AI market. Recommended read:
References :
Ryan Daws@AI News
//
OpenAI has secured a monumental $40 billion funding round, spearheaded by SoftBank, elevating the company's valuation to $300 billion. This substantial financial backing marks one of the most significant capital infusions in the tech industry, underscoring the robust investor confidence in OpenAI's vision and strategic direction. The funding is expected to fuel expanded research and development efforts, upgrades to computational infrastructure, and the accelerated rollout of next-generation AI models. This investment also solidifies OpenAI's position as the world’s second-most valuable private company, trailing only Elon Musk’s SpaceX Corp.
As part of this major announcement, OpenAI revealed plans to release its first "open-weight" language model since 2019. This move signifies a dramatic strategic shift for the company, which has largely built its business on proprietary AI systems. OpenAI CEO Sam Altman confirmed the development on X, stating that the new model would have enhanced reasoning capabilities and be made available in the coming months. This decision comes amid increasing pressure from competitors in the open-source AI space and is driven by the evolving economics of the AI landscape. Recommended read:
References :
Ryan Daws@AI News
//
OpenAI is set to release its first open-weight language model since 2019, marking a strategic shift for the company. This move comes amidst growing competition in the AI landscape, with rivals like DeepSeek and Meta already offering open-source alternatives. Sam Altman, OpenAI's CEO, announced the upcoming model will feature reasoning capabilities and allow developers to run it on their own hardware, departing from OpenAI's traditional cloud-based approach.
This decision follows OpenAI securing a $40 billion funding round, although reports suggest a potential breakdown of $30 billion from SoftBank and $10 billion from Microsoft and venture capital funds. Despite the fresh funding, OpenAI also faces scrutiny over its training data. A recent study by the AI Disclosures Project suggests that OpenAI's GPT-4o model demonstrates "strong recognition" of copyrighted data, potentially accessed without consent. This raises ethical questions about the sources used to train OpenAI's large language models. Recommended read:
References :
@www.eweek.com
//
OpenAI has announced that ChatGPT now has 400 million weekly active users, marking a substantial increase from 300 million in December 2024. This growth of 33% in just three months highlights the platform's increasing popularity across both consumer and enterprise sectors. Brad Lightcap, OpenAI's COO, attributed this surge to word-of-mouth referrals and the perceived utility of the product. In addition to individual users, OpenAI has also seen its enterprise client base double, reaching 2 million paying business users, which includes major companies like Uber, Morgan Stanley, and T-Mobile.
OpenAI's API traffic has also experienced significant growth, doubling in the past six months. Notably, the usage of the "o3" reasoning model has increased fivefold, demonstrating a growing reliance on OpenAI's models in software development and enterprise applications. Despite facing competition from companies like DeepSeek, OpenAI continues to solidify its position as a leading AI provider. Lightcap compared the trajectory of OpenAI's enterprise business to the early growth of cloud services pioneered by Amazon Web Services. Recommended read:
References :
@www.eweek.com
//
OpenAI's ChatGPT has achieved a major milestone, now boasting over 400 million weekly active users. This growth reflects the platform's increasing adoption by both individual consumers and enterprise clients. The number of paying enterprise users has doubled since September, reaching 2 million. This surge is attributed to users recommending ChatGPT to their workplaces, with companies recognizing its potential as a time-saving tool.
OpenAI is also expanding the availability of its "Operator" AI agent, granting access to ChatGPT Pro subscribers in countries like Australia, Brazil, Canada, India, Japan, South Korea, and the United Kingdom. This AI agent can independently operate a web browser and perform tasks for users. Additionally, OpenAI has taken action against malicious activities by removing accounts suspected of being used for surveillance and opinion-influence operations, particularly those originating from China and North Korea. These measures underscore OpenAI's commitment to broadening access while safeguarding against misuse. Recommended read:
References :
@shellypalmer.com
//
References:
shellypalmer.com
, singularityhub.com
OpenAI CEO Sam Altman has outlined the company's roadmap for GPT-5, the successor to GPT-4, with a focus on creating a "magic unified intelligence." This entails developing a single reasoning engine instead of offering multiple AI models like GPT-4 and GPT-4o. Altman suggests that if successful, this approach could significantly improve usability, efficiency, and overall AI intelligence. However, this move has sparked debate within the AI community.
Concerns have been raised about the potential for thought homogenization if GPT-5 becomes the dominant AI for various tasks. The AI community is discussing the shift from multiple AI models to a single reasoning engine and the potential impact on AI development. There's a risk that reliance on a single AI model could limit diversity of thought and stifle innovation, potentially leading to a cognitive monoculture where AI-assisted patterns influence thinking. Recommended read:
References :
Carl Franzen@AI News | VentureBeat
//
OpenAI CEO Sam Altman has recently shared his predictions about the advancements expected in the field of AI, specifically regarding GPT-5 and the next two years of AI development. During a panel discussion, Altman stated that the progress from February 2025 to February 2027 will be even more impressive than the advancements of the last two years. He expressed strong confidence in AI's potential to accelerate scientific discovery, predicting AI systems will compress 10 years of scientific progress into a single year, potentially leading to breakthroughs in climate change and disease treatment.
Altman also made a point of explicitly referencing GPT-5 and its capabilities. Altman posted on X that OpenAI is working toward a "magic unified intelligence," a single reasoning engine, rather than multiple AI models. No more choosing between GPT-4, GPT-4o, o3-mini, or any other variant. One model to rule them all. In a recent blog post, Altman outlined three observations about the economics of AI, and warned that AI could lead to economic inequality. He suggested exploring ideas like giving everyone a 'compute budget' to use AI or relentlessly driving down the cost of intelligence. Recommended read:
References :
|
BenchmarksBlogsResearch Tools |