When OpenAI announced its o1 models in September, the tech world sat up. These aren’t just upgraded chatbots—they were a paradigm shift in Generative AI. Designed to think deeper, reason smarter, and solve puzzles that make even experts sweat, the o1 models mark a significant leap in the continuous evolution of GPT technology.

So, what makes them special? Imagine a model that doesn’t just spit out answers but plans, verifies, and solves problems like a top-tier researcher. OpenAI is delivering on its promise to move AI beyond surface-level interactions, and its latest version of o1 is its boldest statement yet.

What You Need to Know

As Sam Altman noted during the announcement, “The o1 model is the culmination of years of research into making AI not just faster, but more capable of solving real-world problems.”  

Now publicly available through ChatGPT Pro, the full version of o1 delivers up to 2x faster compute times for complex tasks while being 50% more cost-efficient per token than earlier iterations.  

OpenAI’s engineers also indicate how o1 can excel in tasks like multi-step reasoning and data synthesis during the live stream, showcasing its ability to analyse intricate datasets in seconds. For instance, o1 can synthesise climate data spanning decades to predict rainfall patterns — a task previously requiring teams of specialists and weeks of computing time.

Beyond raw numbers, OpenAI also revealed a new pricing strategy. By bundling o1 access into the USD 200 monthly ChatGPT Pro subscription, OpenAI balances accessibility with substantial computing resources required to power the model.

The o1 models aren’t just technical achievements: they’re practical, with applications ranging from simplifying education to diagnosing rare medical conditions.

Why This Is a Big Deal

AI that reasons? That’s a game changer. OpenAI’s o1 models unlock new possibilities in industries like healthcare, education, and programming.

• In Healthcare: They analyse medical patient data to enhance diagnostic accuracy and identify root causes of symptoms.

• In Education: They break down dense scientific topics like quantum mechanics and molecular physics into easy, step-by-step explanations.

• In Programming: They debug, optimise and document code like a senior developer.

By making AI smarter and more intentional, o1 models bridge the gap between human expertise and machine intelligence.  

The Bigger Picture: A Growing AI Arms Race

However, OpenAI doesn’t have the field to itself. Alibaba is throwing its hat into the ring with QwQ-32B-Preview -, a next-gen reasoning model boasting 32.5 billion parameters. Currently, Alibaba’s model can handle prompts up to 32,000 words, rivalling o1 in complex logic and math tasks.

Features of QwQ-32B-Preview:

• Open source under the Apache 2.0 license allows developers complete freedom to experiment.

• Strong performance on reasoning benchmarks like AIME and MATH tests.

However, it’s not perfect. Early reviews highlight language-switching hiccups and occasional looping errors.

Meanwhile, Google is also upping its game in reasoning-based AI, expanding computing power and research teams to refine their models. The race to dominate AI reasoning is heating up, and we’re only at the starting line.

Where ChatGPT Pro Fits In

If you’re wondering how to get your hands on these cutting-edge o1 models, look no further than ChatGPT Pro. Available for USD 200 per month, the recently announced Pro plan provides full access to o1’s high-compute functionality built for researchers, developers, and educators who handle intricate datasets or demanding queries.

OpenAI also launched the ChatGPT Pro Grant Program, offering free o1 access to medical and climate science researchers. It’s a thoughtful move to showcase AI’s societal benefits while ensuring these advanced tools reach those who need them most.

Test-Time Compute: The Secret Sauce

Test-time computing is what sets models like o1 and QwQ apart. As a capability that enables extra processing time during inference, this ability to simulate deliberate thought is why these models excel in complex, nuanced tasks.

What Comes Next?

Since their launch in September 2024, OpenAI’s o1 models have gained rapid momentum, offered faster processing and reduced significant errors by 34%. OpenAI o1 models continue to make medical imaging and engineering strides with new features like advanced image analysis.  

However, with a USD 200/month price tag for o1 Pro via ChatGPT Pro, is it worth the cost compared to its USD 20/month competitor, Anthropic’s Claude Sonnet 3.5?

Key Findings:

o1 Pro excels in complex reasoning, PhD-level mathematics, and vision analysis, making it ideal for specialised tasks requiring deep technical precision. However, it’s slower, often taking 20–30 seconds longer per response.  

Claude Sonnet 3.5 delivers faster, more consistent results, such as coding tasks with cleaner, more practical solutions and handling 95% of real-world scenarios effectively.  

Both models tie in scientific reasoning, with o1 Pro offering deeper analysis and Claude providing clearer, actionable insights.

Verdict:

‍For advanced vision or niche tasks, o1 Pro is worth the investment. For most users, Claude Sonnet 3.5 offers superior value, speed, and versatility at a fraction of the cost. Choose based on your needs!

The Final Word

OpenAI’s o1 models are more than a technical upgrade—they’re a statement about AI’s future. However, the competition is fierce as Alibaba’s QwQ and Google’s projects enter the fray.

So, is o1 the reasoning champion we’ve been waiting for? Or will open-source alternatives like QwQ tip the scales? One thing’s for sure: the AI landscape has never been more exciting.

Ready to experience the future of reasoning? Try ChatGPT Pro today—because the next big breakthrough is just a few computations away.

‍