[ad_1]
An anonymous insider has leaked intriguing details about OpenAI’s upcoming GPT-4.5 language model, shedding light on the AI system’s advanced capabilities and substantial price tag. Although OpenAI CEO Sam Altman denies the validity of the leak, the information revealed offers a glimpse into the potential next phase of large language models.
Overview of leaked GPT-4.5 details
On December 14, an anonymous Reddit user posted apparently leaked prices and details about GPT-4.5, the successor to the popular GPT-3 language model that powers chatbots like ChatGPT. The leaker claims that GPT-4.5 will include:
- Multimodal capabilities for text, images, video, audio and 3D modeling
- Complex reasoning and cross-modal understanding
- Significantly higher prices than GPT-3 for input and output tokens
Key details about GPT-4.5’s alleged pricing:
| Fashion model | Entry price (per 1k tokens) | Output price (per 1k tokens) |
|---|---|---|
| GPT-4.5 | $0.06 | $0.18 |
| GPT-4.5 64k | $0.12 | $0.36 |
| GPT-4.5 Audio/Speech | $0.012 per minute | $0.024 per minute |
If accurate, these prices show a significant increase over GPT-3, which currently costs $0.002 per 1k tokens. The leaked information also hints at a launch in the near future, although OpenAI has not yet officially announced GPT-4.5.
CEO of OpenAI disputes the validity of the leak
When asked directly on social media if the GPT-4.5 leaks were real, OpenAI CEO Sam Altman simply replied “nah.” This matches Reddit commenters’ skepticism about the veracity of the leaked prices and details.
Without confirmation from OpenAI, it is wise to wonder whether an anonymous internet user actually has prior knowledge of the development of GPT-4.5. The lack of details about the model architecture or dataset raises doubts.
However, it is also likely that an actual OpenAI team member leaked accurate information. Similar price leaks preceded GPT-3’s launch in 2020. Regardless, Altman’s outright denial keeps official details about GPT-4.5 secret.
Consequences if the leak is valid
While maintaining a healthy skepticism is wise, the possibility remains that the GPT-4.5 leak paints a real picture of OpenAI’s next step. If the details revealed prove accurate, they have intriguing implications.
Most notable are the claims that GPT-4.5 handles modalities such as audio, video, and images. Integrating vision and speech understanding with language would be a major advance for AI assistants.
The alleged price increase also indicates that OpenAI expects much higher demand for GPT-4.5 compared to GPT-3. Combined with the more advanced capabilities, this suggests a significantly larger underlying model scale.
In fact, the leaked prices resemble some experts’ predictions about the enormous computing power that would be needed to achieve human-level intelligence. The high cost could indicate that GPT-4.5 is approaching this threshold in important areas such as reasoning.
Comparison with potential competitors
If valid, the information leaked on GPT-4.5 illuminates OpenAI’s strategy against AI rivals such as Google and Anthropic. Priced as disclosed, GPT-4.5 would offer significant advantages over GPT-3 in breadth and reasoning ability.
Anthropic has also hinted at a multimodal language model called Claude. But with no official timelines or details from Anthropic, it’s unclear whether Claude will soon be able to match GPT-4.5’s capabilities at competitive prices.
Most intriguing is how GPT-4.5 might compare to models within Google. Leading AI scientist Silver stated in 2021 that Google’s language models are the most advanced made public. If the leak is valid, GPT-4.5 is likely intended to catch up or surpass Google’s internal systems.
Next steps in tracking release
For those eager for the next evolution of language AI, the open questions now are if and when GPT-4.5 can be formally launched. While Altman dismissed the leaks, OpenAI could always have a surprising revelation in store soon.
Enthusiasts speculate based on the rollout of GPT-3 that OpenAI could first grant limited beta access to select partners and users. This could reveal GPT-4.5’s capabilities more concretely before a gradual public debut.
Regardless, the AI community will be watching closely to see if there are any demonstrations that match the versatility the leak claimed. Developer conferences and publications from OpenAI scientists are excellent channels for monitoring. Any indication that multimodal systems are in the works would further fuel the excitement.
The leaked data itself also warrants tracking to detect changes. A changing price chart could reveal updates to the underlying GPT-4.5 model in response to newer benchmarks. If you see the numbers evolving, the ultimate launch could be in the forefront.
Conclusion: leakage stimulates anticipation
While we are fairly cautious about unverified rumors, the supposed inside information about GPT-4.5 is intriguing precisely because the model described would represent such a monumental leap. The leak could put healthy pressure on OpenAI to quickly realize this speculative potential, even if the company is currently downplaying the claims.
If anything, the enthusiastic response to this anonymously posted information on social media shows the anticipation for what the next step of AI could be. When OpenAI finally puts the finishing touches to its creation, whether it’s an incremental upgrade or a revolution, the spotlight will shine brighter than ever on its move toward the future. Every hint of what looms on the horizon has the power to captivate entire communities following this history.
Do you have burning questions about GPT 4.5? Do you need some extra help with AI tools or something else?
Feel free to send an email to Arva, our expert at OpenAIMaster. Send your questions to support@openaimaster.com and Arva will be happy to help you!