The New York Times is reporting that OpenAI has secured $8.3 billion in a new round of funding, months ahead of schedule.

According to the NYT report, this round of investors, which includes Blackstone, TPG, T. Rowe Price, Fidelity, Andreessen Horowitz, and other financial giants, has helped blow up OpenAI’s valuation to $300 billion.

OpenAI has been planning to raise $40 billion by the end of 2025. When OpenAI announced its ambitions back in March, SoftBank immediately provided $30 billion towards the goal.

Now, in addition to $2.5 billion from venture capitalists earlier this year, OpenAI has raised a reported $40.8 billion, way ahead of schedule.

For its latest round, OpenAI’s biggest investor was Dragoneer Investment Group, a venture capitalist firm that coughed up $2.8 billion for the leading AI project.

Furthermore, OpenAI’s revenues are up $3 billion over the last month, with paid subscribers reaching the 5 million mark.

Co-founder Sam Altman’s other project, the ID-focused crypto Worldcoin (WLD), has also been growing this year, launching in the US in May.

Despite the expansion of iris-scanning devices, WLD is dipping alongside the rest of the crypto markets, trading for $0.987 at time of writing, about even with its position three months ago, and down 52% since last year.

Follow us on X, Facebook and Telegram

Don’t Miss a Beat – Subscribe to get email alerts delivered directly to your inbox

Check Price Action

Surf The Daily Hodl Mix

&nbsp

Disclaimer: Opinions expressed at The Daily Hodl are not investment advice. Investors should do their due diligence before making any high-risk investments in Bitcoin, cryptocurrency or digital assets. Please be advised that your transfers and trades are at your own risk, and any losses you may incur are your responsibility. The Daily Hodl does not recommend the buying or selling of any cryptocurrencies or digital assets, nor is The Daily Hodl an investment advisor. Please note that The Daily Hodl participates in affiliate marketing.

Generated Image: DALLE3

OpenAI’s new “o3” language model achieved an IQ score of 136 on a public Mensa Norway intelligence test, exceeding the threshold for entry into the country’s Mensa chapter for the first time.

The score, calculated from a seven-run rolling average, places the model above approximately 98 percent of the human population, according to a standardized bell-curve IQ distribution used in the benchmarking.

The finding, disclosed through data from independent platform TrackingAI.org, reinforces the pattern of closed-source, proprietary models outperforming open-source counterparts in controlled cognitive evaluations.

O-series Dominance and Benchmarking Methodology

The “o3” model was released this week and is a part of the “o-series” of large language models, accounting for most top-tier rankings across both test types evaluated by TrackingAI.

The two benchmark formats included a proprietary “Offline Test” curated by TrackingAI.org and a publicly available Mensa Norway test, both scored against a human mean of 100.

While “o3” posted a 116 on the Offline evaluation, it saw a 20-point boost on the Mensa test, suggesting either enhanced compatibility with the latter’s structure or data-related confounds such as prompt familiarity.

The Offline Test included 100 pattern-recognition questions designed to avoid anything that might have appeared in the data used to train AI models.

Both assessments report each model’s result as an average across the seven most recent completions, but no standard deviation or confidence intervals were released alongside the final scores.

The absence of methodological transparency, particularly around prompting strategies and scoring scale conversion, limits reproducibility and interpretability.

Methodology of testing

TrackingAI.org states that it compiles its data by administering a standardized prompt format designed to ensure broad AI compliance while minimizing interpretive ambiguity.

Each language model is presented with a statement followed by four Likert-style response options, Strongly Disagree, Disagree, Agree, Strongly Agree, and is instructed to select one while justifying its choice in two to five sentences.

Responses must be clearly formatted, typically enclosed in bold or asterisks. If a model refuses to answer, the prompt is repeated up to ten times.

The most recent successful response is then recorded for scoring purposes, with refusal events noted separately.

This methodology, refined through repeated calibration across models, aims to provide consistency in comparative assessments while documenting non-responsiveness as a data point in itself.

Performance spread across model types

The Mensa Norway test sharpened the delineation between the truly frontier models, with the o3’s 136 IQ marking a clear lead over the next highest entry.

In contrast, other popular models like GPT-4o scored considerably lower, landing at 95 on Mensa and 64 on Offline, emphasizing the performance gap between this week’s “o3” release and other top models.

Among open-source submissions, Meta’s Llama 4 Maverick was the highest-ranked, posting a 106 IQ on Mensa and 97 on the Offline benchmark.

Most Apache-licensed entries fell within the 60–90 range, reinforcing the current limitations of community-built architectures relative to corporate-backed research pipelines.

Multimodal models see reduced scores and limitations of testing

Notably, models specifically designed to incorporate image input capabilities consistently underperformed their text-only versions. For instance, OpenAI’s “o1 Pro” scored 107 on the Offline test in its text configuration but dropped to 97 in its vision-enabled version.

The discrepancy was more pronounced on the Mensa test, where the text-only variant achieved 122 compared to 86 for the visual version. This suggests that some methods of multimodal pretraining may introduce reasoning inefficiencies that remain unresolved at present.

However, “o3” can also analyze and interpret images to a very high standard, much better than its predecessors, breaking this trend.

Ultimately, IQ benchmarks provide a narrow window into a model’s reasoning capability, with short-context pattern matching offering only limited insights into broader cognitive behavior such as multi-turn reasoning, planning, or factual accuracy.

Additionally, machine test-taking conditions, such as instant access to full prompts and unlimited processing speed, further blur comparisons to human cognition.

The degree to which high IQ scores on structured tests translate to real-world language model performance remains uncertain.

As TrackingAI.org’s researchers acknowledge, even their attempts to avoid training-set leakage do not entirely preclude the possibility of indirect exposure or format generalization, particularly given the lack of transparency around training datasets and fine-tuning procedures for proprietary models.

Independent Evaluators Fill Transparency Gap

Organizations such as LM-Eval, GPTZero, and MLCommons are increasingly relied upon to provide third-party assessments as model developers continue to limit disclosures about internal architectures and training methods.

These “shadow evaluations” are shaping the emerging norms of large language model testing, especially in light of the opaque and often fragmented disclosures from leading AI firms.

OpenAI’s o-series holds a commanding position in this testing workflow, though the long-term implications for general intelligence, agentic behavior, or ethical deployment remain to be addressed in more domain-relevant trials. The IQ scores, while provocative, serve more as signals of short-context proficiency than a definitive indicator of broader capabilities.

Per TrackingAI.org, additional analysis on format-based performance spreads and evaluation reliability will be necessary to clarify the validity of current benchmarks.

With model releases accelerating and independent testing growing in sophistication, comparative metrics may continue to evolve in both format and interpretation.

Mentioned in this article

OpenAI is working on yet another AI model reportedly called GPT-4.1, a successor to GPT-4o, which is expected to come before GPT 5.0

The Verge recently reported that OpenAI plans to launch GPT-4.1, which is an upgrade to the existing GPT-4o. And now, we have more reasons to believe the model is indeed coming.

As spotted by AI researcher Tibor Blaho, OpenAI is already testing model art for o3, o4-mini, and GPT-4.1 (including nano and mini variants) on the OpenAI API platform.

This confirms that GPT-4.1 does exist, but it doesn’t appear to be a successor to GPT-4.5.

My understanding is that GPT-4.1 is a successor to GPT-4o, which is multimodal. On the other hand, GPT-4.5 focuses more on creativity and delivering better answers.

In a “Pre-Training GPT-4.5” video from OpenAI, founder and CEO Sam Altman dropped hints that OpenAI has a team that wants to redo GPT-4 from scratch using new training data and systems.

“If you guys could go pick whoever you wanted, what is the smallest team from OpenAI that could go retrain GPT-4 from scratch today with everything we know and have and all the systems work?” Sam said.

It’s unclear if Altman is referring to the new GPT-4.1 model, but if I were to bet, I’d bet on GPT-4.1.

Also, GPT-5 isn’t happening anytime soon, as OpenAI plans to focus on o3, o4-mini, o4-mini-high, and GPT-4.1 (including nano and mini variants).

The following is a guest post from Ahmad Shadid, Founder of O.xyz.

In an era where artificial intelligence dictates global power dynamics, OpenAI is making bold moves to secure America’s dominance in the sector. Its new plan called the “AI Action Plan,” tries to ease the regulatory framework, implement export controls, and increase federal investment to stay ahead of China’s AI expansion. 

OpenAI signed a deal with the Trump administration for the new “AI Action Plan” on March 13. The deal revolved around limited regulatory oversight and rapid development of AI in the US. 

The proposal highlights a fundamental truth: too much state-level regulation might undermine America’s leadership in AI, even as China’s state-supported AI players — led by DeepSeek — continue to grow rapidly. 

Defending AI from censorship

DeepSeek’s R1 model, released in January 2025, performed at the level of top US AI systems — though having been developed at a much lower cost — challenging the American tech giants’ dominance. 

This proved to be a major sell-off in US tech stocks, with firms such as Nvidia suffering huge losses. Soon after, the US government raised red flags about national security and data privacy, debating on policy solutions to keep America out front in the very technologies it wrote the rules for.

OpenAI’s approach represents a pivotal point in American AI policy, combining regulatory advocacy with industrial ambition to ensure the US stays on top of the game when it comes to AI. Moreover, at the heart of OpenAI’s plan is an export-control strategy that aims to limit the country’s expanding influence in China. 

This will prevent the misuse of AI platforms and technologies by opposing nations. Consequently, the export controls will protect the US national security.

OpenAI’s plan also calls for using federal dollars to explain to the world that American-made AI is safer and that US-based companies should stay ahead of the international AI stream. 

DeepSeek is not only a Chinese AI initiative and a commercial competitor but also a fundamental ally of the Chinese Communist Party (CCP). In late January, DeepSeek grew infamous for blocking information on the 1989 Tiananmen Square massacre, riding a wave of screenshots on social media pointing out China’s censorship. 

The $500 billion plan

A central part of OpenAI’s pitch is locking in greater federal funding for AI infrastructure. This means ensuring that the high-water mark for American progress in the field of AI does not merely concern protecting what comes next from foreign threats but also reinforces the necessary computational and data infrastructure to sustain long-term growth. 

The Stargate Project, for instance, is a joint effort by OpenAI, SoftBank, Oracle, and MGX that will provide up to $500 billion for the development of AI infrastructure in the United States. 

This ambitious initiative is intended to cement American AI superiority while producing thousands of domestic jobs — much against the belief of the ‘AI might replace your job’ narrative.

It is a major tactical shift in the approach to AI policy, acknowledging that private sector investments are not enough to remain competitive compared to state-sponsored efforts like China’s DeepSeek. 

The Stargate Project wants to ensure the construction of advanced data centers and the expansion of semiconductor manufacturing within the United States to keep AI development domestic in the US. 

In its early stages, Federal support for AI infrastructure is critical — both for growing economic competitiveness and for national security. AI-powered features are often used in national defense and intelligence. For instance, Shield AI’s Nova is an autonomous quadcopter drone that uses AI to fly itself through complex environments without GPS to gather life-saving information in combat environments. 

Furthermore, AI is also critical in cyber defense against hacking, phishing, ransomware, and other cybersecurity threats because it can identify deviations or abnormalities in systems in real time. Hence, its role in detecting patterns and spotting irregularities helps the US safeguard critical defense infrastructure from cyberattacks, making it all the more important to fast-track AI’s growth for defense.

Battle for AI training models

A key element of OpenAI’s proposal is the call for a new copyright approach that would ensure that American AI models can access copyrighted material for use in their training. The ability to train on a wide range of datasets is vital to keeping AI models sophisticated.

If the copyright policies are restrictive, it could put the US at a disadvantage to their foreign competitors — especially the Chinese ones, which operate among weaker copyright enforcement.

AI tools are evaluated for risk, governance board scrutiny, and compliance verification frameworks such as the House AI Policy and DHS conditional approvals. Although FedRAMP’s ‘fast pass’ may expedite deployment, antennae from the FTC and regulations will keep AI’s purposes on the same shelf as national security policy and consumer protections. 

These safeguards, while no doubt very important, often slow the pace of AI adoption in crucial government use cases.

Now, OpenAI, in particular, is lobbying for a partnership between the government and the industry, where AI companies voluntarily contribute their models’ data, and in exchange, they would not be liable to strong state restrictions. 

Not an easy road ahead

Though OpenAI’s proposal is bold and ambitious, it poses serious questions about how much regulation can help in driving innovation in this burgeoning sector. 

While weakening state-level regulatory oversight will leave room for faster AI developments, there is a significant concern yet to be resolved. The partnership structure of AI organizations with the federal government could create the potential for private firms to exercise outsized power over the nation’s policies on AI and the users themselves. 

Regardless of these fears, one thing is clear: the United States cannot afford to fall behind its competitors in AI development. If done right, this partnership could guarantee that American AI stays the dominant framework worldwide rather than yielding ground to foreign government-controlled competitors like China’s DeepSeek.

Mentioned in this article