Source: Alphabet List
OpenAI's 12-day marathon live broadcast event has ended, but this "marathon" is a bit unworthy of its name.
The most anticipated GPT-5 is still absent, and the belated Sora lacks surprises. The next-generation reasoning model o3 has not yet been released, and it will have to wait until January next year to eat it.
At the same time, external pressure has come one after another: Google took advantage of OpenAI's marathon event to raise a sniper rifle and fired one bullet after another that hit the target accurately. Musk's xAI announced the completion of a $6 billion Series C financing not long after the OpenAI event ended. Together with the Series B financing in May, the company has raised $12 billion.
This was originally a marketing opportunity for OpenAI to end 2024 in a relaxed and cheerful atmosphere, but it formed a reflection with the external pinch, perfectly outlining OpenAI's experience this year.
In this year, OpenAI is still excellent, but perhaps it has retreated from the altar to the human world. And 2025 is destined to have more challenges.
At the end of the year, before Christmas, a series of new products and features were launched in one go with 12 consecutive days of live broadcasts!
The above is the impression before the start of OpenAI's year-end live broadcast event. But in fact, the 12-day live broadcast is not continuous, and it is off on weekends; the live broadcast days are not long, sometimes only ten minutes.
The most important thing is that in terms of content, OpenAI did announce a series of new features and products, especially the video generation tool Sora was finally launched, and the next-generation reasoning models o3 and o3 mini were launched.
However, the most anticipated GPT-5 by the outside world is still absent.
Just after the 12-day live broadcast event officially ended, the Wall Street Journal broke the news that the development of GPT-5 (codenamed Orion) was slow and did not achieve the expected results.
Specifically, the report said that the Orion project has been developed for 18 months and at least two large-scale trainings have been conducted, each of which took months to process large amounts of data to make Orion smarter. But relevant people said that new problems will arise every time training is carried out.
Even with Orion's best performance, although its performance is better than OpenAI's existing model, it is not worth it when combined with the cost-according to the Wall Street Journal, the computing cost of six months of training alone may be as high as $500 million.
Of course, this situation also made Microsoft, OpenAI's largest funder, "the first to be disappointed." Microsoft's original expectation was that Orion would see a new model in mid-2024.
Looking back at the end of November 2022, ChatGPT was just released, and GPT-4 was launched in March of the following year, and it was still a crushing existence. Soon after, in mid-2023, the development of the GPT-5 project was launched, and from then on, the outside world was looking forward to the launch of GPT-5.
The greater the expectations, the greater the disappointment. The obstruction of the launch of GPT-5 has also become a conspicuous sign of OpenAI's transformation in the eyes of the outside world.
While GPT-5 was delayed, OpenAI also made other efforts, such as launching the inference model o1, and then launched o3 in this live broadcast event, but it lacked the surprise of GPT-4 when it was first launched.
In addition to the model changes behind ChatGPT, OpenAI's Sora was officially announced at the beginning of the year and finally launched during the live broadcast event at the end of the year. One year is indeed a long time in the booming AI industry. "Friendly competitors" have already laid out their plans, and Sora is no longer "condescending".
2024 has become a key year for OpenAI's image transformation. It cannot be said that OpenAI is no longer excellent, but at least it has fallen from the altar, and the "crushing" advantage seems to be gone.
"Friendly competitors" have risen up, making it increasingly difficult for OpenAI to "stay far ahead".
Just for this live broadcast event, OpenAI's most popular o3 model and Sora were blocked. The most powerful competitor is Google, which was once controversial for its slow action in the AI wave.
On December 9, the third day of the live broadcast, OpenAI finally announced the official launch of Sora, which is available to ChatGPT Pro users who pay $200 per month.
Unlike the high attention caused by Sora's official announcement at the beginning of the year, this time Sora is like a pebble dropped into a lake and did not cause much waves. The performance of up to 20 seconds and the highest picture quality of 1080p is really limited in appeal. After all, the biggest competitor Runway and domestic Keling and Conch AI have been launched for a long time, and the price is not that expensive.
On December 17, Google came to "add insult to injury" again and suddenly launched Veo2.Veo2 can currently create up to 8 seconds, 720p videos, and promises to reach more than 2 minutes and 4K resolution in the future. CEO Sundar Pichai emphasized on social media that Veo2 "has a better understanding of real-world physics and motion", which clearly shows who he was mocking.
Now, the limelight has been completely taken away. After Sora was launched, users' discussions were not enthusiastic. Many reviews also pointed out that it still had the biological motion deformation, unnatural hands, and garbled text in the demonstration at the beginning of the year. Veo, on the other hand, has attracted a lot of praise, especially for its coherent biological motion, natural and rich human expressions, and stable long-panning shots.
Under pressure, as the live broadcast ended, OpenAI announced that it would provide unlimited access to Sora to all subscribers during the Christmas holiday.
Google's blockade is not limited to this.
On December 20, the penultimate day of OpenAI's "12-day marathon" live broadcast event, Google announced the launch of Gemini 2.0 Flash Thinking. This is a multimodal reasoning model that visualizes the thinking chain process and highlights the transparency of AI thinking and fast problem solving.
Friends familiar with OpenAI products should not find it difficult to see that Google's new model is aimed at OpenAI's reasoning model o1.
According to the preliminary evaluation results of the independent benchmarking website lmarena.ai, Gemini 2.0 Flash Thinking's overall performance surpassed the o1 preview version and ranked first in total score.
In addition, Google's actions in AI at the end of the year also include: releasing a new generation of literary graph model Imagen 3, and integrating the previously released Astra project and Mariner project into Gemini2.0.
Since the launch of GPT-4, Google has been constantly reorganizing internally to concentrate its AI firepower. In April last year, Google merged Google Brain and DeepMind into "Google DeepMind", which was headed by DeepMind co-founder Demis Hassabis. In August of this year, Google absorbed Character.AI's founder Noam Shazeer and "a small number of colleagues" into Google DeepMind.
Today, Google has swept away the embarrassment of "not keeping up" at the beginning of last year and has become one of OpenAI's biggest competitors. According to data from the unified LLM API service platform OpenRouter, Google's share of platform developers has grown to 50%, while this September the figure was only about 5%.
Google is just one of the "friendly competitors" that have risen up. The most representative one is Anthropic, which started as an OpenAI "rebel". It released Claude 3.5 this year, with the support of Amazon. xAI, founded by OpenAI's "enemy" Musk, released Grok-2 this year, and changed from being open only to X subscribers to being open to all users. It is also rumored that it will launch a separate application. And the giant Meta continued to consolidate its "open source AI" strategy this year and continued to increase its supercomputing deployment.
Menlo Ventures surveyed 600 IT decision makers in US companies and released the results of the survey in 2024: this year, corporate spending on generative artificial intelligence has soared 500%, from $2.3 billion in 2023 to $13.8 billion.
In the field of enterprise artificial intelligence, OpenAI's market share has dropped from 50% to 34%, Anthropic has doubled from 12% to 24%, Meta's share has remained at 16%, and Google has also increased significantly from 7% to 12%.
In 2024, OpenAI's competitors are getting stronger and more fanged, making OpenAI surrounded by layers.
Just talking about the competition in products is not enough to see the full picture of OpenAI's situation.
Competition in the AI industry is not static. Under the manifestation of OpenAI's weak products is the company's fierce structural transformation and personnel turmoil.
During the OpenAI live event, another key figure left, Alec Radford.
He has been working at OpenAI for 8 years. His idea of combining the Transformer architecture with massive data completely changed OpenAI's research and directly contributed to the success of the later GPT model.
In fact, throughout 2024, OpenAI's personnel was turbulent, and at least 9 senior executives left. There are three points that best illustrate the "turbulence":
First, the 11 people in the OpenAI startup team have been sharply reduced to the current 2 people. Second, former chief scientist Ilya Sutskever and former chief technology officer Mira Murati left. Third, search director Shivakumar Venkataraman left. He previously led Google's search advertising team and was hired by OpenAI to lead the search sector 7 months ago.
Old veterans jumped ship one after another, and newly recruited key figures also left at the speed of light, which shows the degree of turmoil in OpenAI. Most of the powerful figures who left OpenAI, except for a few who started their own businesses, were absorbed by "friendly competitors".
When this wave of AI was first set off, the talent scramble was already on. OpenAI has "feelings and ideals" in its left hand and high salaries in its right hand, becoming a "holy land" that aspiring researchers yearn for. However, whether the magic of OpenAI can continue now is a question mark.
OpenAI's next important transformation is to become a truly profitable company and get rid of the "control" of the non-profit board of directors. This matter was just a rumor last year, but it has been put on the table by OpenAI this year. Embracing profitability will inevitably give up some uniqueness, just as Musk has always attacked: OpenAI was founded as a non-profit organization with the original intention of fighting against technology giants represented by Google and creating AGI that benefits mankind.
The process of restructuring is actually a readjustment of the company's priorities and primary goals, and this process will inevitably cause friction with the "veterans".
High salaries may not last forever.
OpenAI itself does not have a stable blood-making ability. This year, OpenAI completed a $6.6 billion financing. According to the financial documents disclosed by the company, it is expected to make a profit by 2029, when its revenue will reach $100 billion. According to media analysis that has read the documents, including The Information, OpenAI's losses next year may reach $14 billion, nearly three times the expected losses this year.
As for the largest "gold owner" Microsoft, its relationship with OpenAI has become increasingly delicate. On December 24, local time, Reuters quoted people familiar with the matter as saying that Microsoft is working hard to help Copilot get rid of its dependence on OpenAI. In addition to training its own small models, it is also actively customizing other third-party models. Some media simply summarized it as: Microsoft hopes to establish an "open relationship" with OpenAI.
On the other hand, Musk also targeted and attacked OpenAI's high-salary strategy.
In November, Musk's lawsuit against OpenAI escalated. This time, not only new defendants were added, but also new evidence was submitted. In a revised 107-page complaint, the high-salary poaching behavior of OpenAI was described as follows: "OpenAI tried to make its competitors short of AI talent through aggressive recruitment and high salaries. In addition, OpenAI plans to spend $1.5 billion on 1,500 employees."
Musk's resumption of the lawsuit against OpenAI is also worth pondering. The lawsuit first occurred in March this year and was subsequently withdrawn. A few months later, Musk re-filed the lawsuit in federal court. In mid-November this year, the scope of the complaint was further expanded.
At the same time, Musk himself and xAI in his hands are advancing by leaps and bounds. He himself has become one of the biggest beneficiaries of the US presidential election and a "confidant" of Trump, who will be in charge of a brand new "efficiency department" after Trump takes office.
On December 24, xAI officially announced the completion of a $6 billion Series C financing and announced the investment lineup, including Nvidia, AMD, Morgan Stanley, and Sequoia Capital. Together with the $6 billion Series B financing in May this year, xAI's total financing amount has exceeded $12 billion. According to CNBC, xAI's target valuation is $50 billion.
If OpenAI CEO Altman could still laugh it off last year and remain elegant in the face of Musk's attack, then in 2024, this elegance is no longer there.
During this year's financing, Altman tried to reach a closed agreement with investors, urging investors not to invest in OpenAI's competitors.
A few days ago, Altman called Musk "clearly a bully" in an interview and said that Musk's high-profile dispute with OpenAI has become a "sideshow." It is rare for Altman to "attack" Musk so bluntly in public.
At the end of the year, OpenAI may have hoped to end 2024 in a relatively relaxed atmosphere by holding a live broadcast marathon event during Christmas, and to draw the attention of the outside world back to the product itself.
But with competitors attacking and challenges ahead, such efforts do not seem to meet expectations. Instead, people see the pressure on OpenAI and see that a challenging 2025 is roaring towards OpenAI.