OpenAI CEO Sam Altman calls new OpenAI o1 model flawed and limited: Here is why

OpenAI has recently introduced its new AI model with advanced reasoning skills. OpenAI o1, which is likely the codenamed project Strawberry, is designed to solve complex tasks involving maths, coding and other complicated problems. The company claims that this model will also be able to give answers to the PHD-level scientific questions with future updates. OpenAI’s latest offering, the o1, seems to be a remarkable achievement, building upon the groundbreaking success of GPT 4o. The rapid development and release of this new model demonstrates the company’s commitment to continuously improving and innovating, with the goal of further enhancing human productivity and capabilities. Meanwhile, talking about the new model, OpenAI CEO Sam Altman has described it as “still flawed, still limited”.
Altman emphasised that since this is the first version of the new model, there are several limitations to it. The flaws are not because the existing features do not work, but because the new model lacks a few. For instance, there is no image support, it is a little slow to respond and more. Let’s look at it in detail.
As an early model, it doesn’t have features like browsing the web for information and uploading files and images. This means that, for now, it can only take text prompts. So, if you think you can just click a picture, upload it here, and it solves the problem, it can’t do that right now. You have to write down the math problem and only then OpenAI o1 will solve it.
OpenAI release has stated that the o1 model is a little slower compared to the other models. The company reasons that since it is designed to “think before answering”, it takes its own sweet time to solve the problem. OpenAI, in the release, stated that it has trained these models to spend more time thinking through problems before it responds, much like a human would. The model uses a “chain of thought” to process queries, similarly to how humans process problems by going through them step-by-step.
The media reports have been comparing the first version of OpenAI o1 to GPT 4o, but is it viable? I don’t think so. Let me take you back to the time when OpenAI released ChatGPT. Launched in 2022, the platform took the AI world by storm. At that moment, it did not have image support, just like the OpenAI o1. Similarly, this is the first version of AI “reasoning” model that has been trained to answer more complex questions. Asking for everything, which OpenAI has already introduced in the advanced GPT version, is pointless. The company also states that in many common cases GPT-4o will be more capable in the near term.
Moreover, introducing image support to ChatGPT also created an easy way for hallucinations to set in. For now, OpenAI o1 is free of it, for a major part. We need to understand that this model aims to target a different set of people. The company also stated, “These enhanced reasoning capabilities may be particularly useful if you’re tackling complex problems in science, coding, math, and similar fields. For example, o1 can be used by healthcare researchers to annotate cell sequencing data, by physicists to generate complicated mathematical formulas needed for quantum optics, and by developers in all fields to build and execute multi-step workflows.”
Look at the graph, representing how the new model works, posted by the CEO:
The OpenAI o1 comes at a much higher price. In the API, o1-preview is $15 per 1 million input tokens, and $60 per 1 million output tokens. For comparison, GPT 4o price is set at $5 for 1 million input tokens and $2.50 for 1 million input token. For India, it means that where GPT 4o is available at around Rs 420 per month for 1 million input token, the users have to pay Rs 1,260 for the same to access OpenAI o1. While the company has also released a cheaper version with it, OpenAI o1 mini, the major updates are expected to come in the original variant.
The model has been rolled out for all ChatGPT Plus and Team users. It will be interesting to see how the Indian audience will react to the price. Will we jump in to try OpenAI o1 or continue to use GPT 4o and wait for it to bring more updates? I bet for the latter option.