Last week we talked about the possible arrival of ChatGPT 5.0 this year. We knew that OpenAI was paving the way for a new update, Project Strawberry, and with that, the rumors began. The wait is over, however, they came with a completely different update: ChatGPT o1, and this news is stirring up the tech world. Let's dive a little into this artificial intelligence revolution!
What is ChatGPT o1?
ChatGPT o1, launched on September 12, 2024, marks a new chapter in the history of AI. OpenAI decided to "reset the counter," naming this version o1, symbolizing a quantum leap in capabilities. It's not just an update; it's a complete reinvention!
The new version, ChatGPT o1, stands out especially in areas of reasoning and creative thinking, being designed to execute longer chains of thought before arriving at a final answer. This ability enhances accuracy in complex tasks, such as in the fields of exact sciences (physics, chemistry, biology), as well as mathematics and programming. In tests, the model was able to solve 83% of the problems from a Mathematics Olympiad, while GPT-4 could only solve 13%, demonstrating the evolution in terms of deep reasoning.
For those who want to understand the explanation behind the version name o1: the "o" refers to OpenAI, and the "1" signifies the reset of the counter.
o1 vs GPT-4: What has changed?
Advanced Reasoning: The o1 uses a hidden "Chain of Thought" (CoT) system, generating deeper and more well-founded responses.
Efficiency: It can handle complex and lengthy tasks without needing to break the context into smaller parts.
Speed: The o1-mini version promises ultra-fast responses, perfect for everyday tasks.
Specialized Knowledge: It excels in STEM fields and programming, surpassing previous versions.
One of the main differences between ChatGPT o1 and GPT-4o is that the new version does not yet offer certain features, such as file uploads and web browsing, which were already available in GPT-4o. This makes GPT-4o still the preferred option for some users, at least for now. However, for those who need an AI optimized for sciences, mathematics, and coding, ChatGPT o1 offers unparalleled performance.
Available Versions
o1-preview: The full version, with all advanced capabilities.
o1-mini: A version optimized for speed, ideal for quick tasks.
The o1-mini is optimized to be faster and more economical, being 80% cheaper than o1-preview, yet still offering excellent reasoning capacity, especially in technical tasks.
Although specific pricing has not yet been disclosed, OpenAI indicates that the cost of o1 is expected to decrease every 1-2 years. For regular users, the expectation is that o1-mini will be available for free in the future. For developers, the o1 model costs $15 per million input tokens, while a million output tokens are worth $60.
ChatGPT Plus and ChatGPT Team users already have access to the o1 version, but with differentiated usage limits: 30 messages per week for o1-preview and 50 messages per week for o1-mini.
Although specific pricing has not yet been disclosed, OpenAI indicates that the cost of o1 is expected to decrease every 1-2 years. For regular users, the expectation is that o1-mini will be available for free in the future. For developers, there are already defined prices, where the o1 model costs $15 per million input tokens, while a million output tokens are worth $60.
Surprising New Features
Philosophical Reasoning: The o1 can discuss deep questions like "What is life?" with a unique perspective.
Enhanced Creativity: It excels in creative tasks, such as poetry and innovative problem-solving.
Generalization Ability: It learns and applies concepts impressively, even with limited instructions.
The Future is Now: What to Expect?
Integrated Tools: Soon, the o1 will support function calls, code interpretation, and web browsing.
Multimodal Capabilities: Integration of text, image, and possibly audio is on the way.
Continuous Improvements: OpenAI promises frequent updates, expanding the model's knowledge and skills.
We have entered a new era of tools, and once again OpenAI is leading the way. What I find remarkable about them is that everything they study and develop is shared with everyone, resulting in many other tools that utilize GPT models in their structures.
ChatGPT o1 is not just a tool; it comes with the proposal of being a thought partner, helping us expand our intellectual and creative horizons. It’s incredible how it can generate complex chains of thought before providing an answer. For those working in STEM (Science, Technology, Engineering, and Mathematics), the o1 is the ideal choice, while GPT-4o still offers better versatility for users needing an AI with broader features, such as web browsing and image generation. As the capabilities of o1 expand, it promises to become an indispensable tool for those seeking advanced AI solutions.
ChatGPT o1 is not just an update; it is a quantum leap in artificial intelligence. We are witnessing the birth of a new era in human-machine interaction. And to conclude, here are some tips on how you can make the most of the o1 version.
Use prompts that encourage detailed reasoning.
Try complex and multifaceted tasks—the o1 loves challenges!
Don't hesitate to ask for detailed explanations—the model was made for that.