In recent developments, OpenAI seems to be gearing up for a significant leap in its offerings with the anticipated release of GPT-4.5. On February 18, OpenAI's CEO, Sam Altman, made a striking statement, emphasizing that the experience of trialing GPT-4.5 by high-demand testers was far deeper than he had expectedThis announcement hints at the model nearing the final stages of testing ahead of a potential release.

Interestingly, Altman's assertive approach seems to be a strategic move against Elon Musk, who unveiled Grok3, his third-generation language model developed by xAI, around the same timeMusk's declaration that Grok3 outperformed existing products during testing captured the attention of both the tech and gaming sectors, fueling excitement around a new AI gaming studio being established by xAI, which in turn spurred market enthusiasm for gaming stocks in A-shares.

OpenAI’s Sudden Announcement

At 1 AM on the 18th, Sam Altman described the experience of using GPT-4.5 in a testing phase as profound for those looking for advanced AI interaction

Advertisements

This suggests that the new model is nearing its official unveiling, stirring considerable anticipation in the tech community.

Recently, Altman also unveiled a roadmap for the highly anticipated GPT-5 model, which is expected to follow in the footsteps of GPT-4, a groundbreaking AI language model that made waves in both tech and policy circles in 2023. In response to inquiries on the X platform, Altman indicated that GPT-5 could be released within months, possibly aiming for a launch in late 2025.

On the X platform, Altman elaborated on the company’s intention to release GPT-4.5, codenamed Orion, within a few weeksThis model will be OpenAI's last non-simulated reasoning model, transitioning to a future that includes more advanced reasoning techniques.

Subsequently, GPT-5 is set to become a system that incorporates features from all of OpenAI’s current AI models, including traditional AI models, as well as specialized models for online searching and research

Advertisements

Altman noted, “We will release GPT-5 in both ChatGPT and API, merging various technologies, indicating that standalone models will no longer be separately published.”

Altman assured that ChatGPT free users would have “unlimited access to GPT-5 dialogues under standard intelligence settings,” while ChatGPT Plus users could operate GPT-5 at higher intelligence levels, and Pro users would have access to the model at the highest intelligence levelHe also mentioned that integrated AI models would utilize multiple ChatGPT extensions, including advanced voice modes, Canvas, search, and in-depth research capabilities.

Musk’s Counter Move

Curiously, Musk also released what he touted as the “most powerful model” Grok3 on the 18th, asserting its superiority.

Reports suggest that Grok3 has outperformed its counterparts in various benchmark tests regarding mathematical reasoning, scientific logic reasoning, and code writing capabilities, making it a formidable contender in the AI landscape.

Musk's foray into AI came with the founding of xAI earlier in 2023, but the development of Grok stretches back to 2015, coinciding with Altman and others founding OpenAI

Advertisements

As time passed, philosophical disparities divided Musk from his OpenAI counterparts, prompting his eventual departure from the organizationThis culminated in the emergence of ChatGPT, an AI chatbot that captivated global interest.

The term "Grok," introduced by author Robert Heinlein in his 1961 novel "Stranger in a Strange Land," referred to an intuitive understanding through empathy, and has since evolved within computer programming cultures.

So, what can we expect from the Grok model? Industry experts note that Grok's performance metrics reveal significant advancements in its problem-solving capabilities, especially in high school competitive benchmarks, showcasing notable generalization abilities over a relatively short 17-month development period.

Moreover, the model’s ability to transfer skills is notable; Grok can tackle varied tasks beyond mathematical reasoning and coding assignments, displaying learning capacities that allow for error detection and correction, thus enhancing its overall competence.

In terms of hardware, the team has assembled GPU clusters and executed various stability tests to enhance the model's reliability

Predictions suggest that AI technologies could even take home prestigious awards like the Fields Medal and Nobel Prize within the next few years.

Moreover, Grok has already demonstrated practical applications, having provided Python code solutions for physical problems related to spacecraft transfers between Earth and MarsWhile the accuracy is yet to be confirmed, the simulation mimics real-life scenarios effectively.

The Power Play: Computing and Energy

On the market front, tech stocks like Sanbian Technology saw significant price surges, attributed to their connection with the xAI concept

As AI competition intensifies, computing power and electricity are emerging as key resources.

Despite advancements like DeepSeek-R1, which significantly reduced training costs and inference expenses, computing power remains crucial for model performanceThe explosion of AI applications is likely to escalate overall demand for computational resources.

Furthermore, emerging trends indicate that the power supply system may also require more attentionRecent reports indicate substantial rises in electric utility stocks like Guodian Power, suggesting a growing demand driven by AI advancementsProjections anticipate a significant increase in summer power peak demands, propelled largely by AI’s continuous evolution.

In conclusion, the developments from OpenAI and xAI signify a vital shift in the AI landscape, promising innovations that could reshape industries

Advertisements

Advertisements