Microsoft is making OpenAI’s o1 reasoning model available for free to all Microsoft Copilot users, eliminating the previous requirement for a paid Copilot Pro subscription.
This means that every user of Microsoft’s AI assistant can now access Think Deeper, a feature that enables structured, multi-step reasoning for more complex queries.
The announcement, made by Microsoft AI CEO Mustafa Suleyman, signals a broader push by Microsoft to integrate OpenAI’s advanced models across its products without additional cost to users.
Suleyman, in a LinkedIn post, emphasized the scale of the rollout, writing, “Today we’ve made Think Deeper free and available for all users of Copilot. This now gives everyone access to OpenAI’s o1 reasoning model in Copilot, everywhere at no cost.”
He also suggested that Microsoft is planning further AI expansions, stating, “We’ve got so much more in the pipeline right now that I can’t wait to tell you about.”
The decision to remove the paywall on Think Deeper sets Microsoft apart from OpenAI, which still charges for direct access to o1 through its ChatGPT subscription tiers.
Until now, users had to subscribe to ChatGPT Plus for $20 per month or opt for the more expensive ChatGPT Pro plan at $200 per month to gain access to o1’s advanced reasoning capabilities.
Microsoft’s move undercuts these plans by providing free access through Copilot, albeit without offering the enhanced o1 Pro Mode, which remains exclusive to OpenAI’s premium subscribers.
Related: OpenAI Unveils New o3 Model With Drastically Improved Reasoning Skills
What Is Think Deeper and How Does It Work?
Think Deeper was initially introduced in October 2024 within Copilot Labs, an experimental environment where Microsoft tests AI features before a broader rollout. The feature is designed to improve the depth and accuracy of responses by applying OpenAI’s o1 model, which specializes in chain-of-thought reasoning.
Unlike standard AI-generated answers that provide quick, often surface-level responses, Think Deeper allows Copilot to process complex queries in multiple steps, delivering more structured and logically consistent answers.
The function is activated by clicking the Think Deeper button within Copilot, prompting the AI to take approximately 30 seconds to analyze a question and generate a well-reasoned response.
Suleyman highlighted practical applications of the feature, explaining, “Plan that epic project. Brain dump everything into Think Deeper and watch it churn through it all and spit out a step by step guide to making it happen. I’ve tried this on a few things (fitness routine, big launch coming up) and it’s genuinely so helpful.”
He also positioned it as a useful tool for research and learning, adding, “Deep dive any topic. Want to learn about something? It nails it! Not a substitute for human teaching, but an extraordinary augmentation. When I get obsessed with something I turn to Think Deeper these days. Lately: the history of India, what’s happening to ocean currents.”
How Think Deeper Compares to Other AI Reasoning Models
Microsoft’s decision to open up Think Deeper for free comes as competition in AI-driven reasoning intensifies. Google recently introduced Gemini 2.0 Flash Thinking, a reasoning model designed to provide transparent, step-by-step problem-solving insights.
Unlike OpenAI’s o1 model, which operates as a black-box AI where users cannot see the internal reasoning steps, Google’s system visually reveals the logical process behind each response. Benchmark results have shown that Gemini 2.0 Flash Thinking outperforms OpenAI’s o1-preview and o1-mini models in instruction-following, creative writing, and longer query tasks.

However, no direct comparison has been made between Gemini 2.0 Flash Thinking and OpenAI’s more powerful o1 Pro Mode.
OpenAI is also advancing its reasoning models with the o3 series. ChatGPT is expected to get o3-Mini soon, a smaller and more efficient variant of the larger o3 model designed for faster response times while maintaining strong problem-solving capabilities.
o3 has demonstrated drastic improvements in logical reasoning over its predecessor o1, scoring 87.5% on the ARC-AGI benchmark, significantly outperforming o1’s best result of 32%. The model also performs much better in the AIME Competition Math benchmark and GPQA Diamond.

OpenAI is positioning o3 as a step toward more adaptable AI systems, capable of handling complex, multi-step tasks more effectively than previous iterations. With the upcoming release of o3-Mini, OpenAI aims to balance speed, efficiency, and reasoning power, potentially expanding its appeal to real-time applications.
By integrating OpenAI’s reasoning models directly into Copilot and offering them for free, Microsoft is expanding its AI ecosystem while differentiating itself from OpenAI’s direct-to-consumer approach.
OpenAI continues to operate a subscription-based model for individual users, requiring payment for its advanced AI capabilities, while Microsoft leverages its partnership with OpenAI to provide those same models as built-in features across its software suite.
Microsoft has made AI a core part of its long-term strategy, embedding OpenAI’s models into Windows 11, Microsoft 365, and Azure AI services. Copilot is a central element of this vision, serving as an AI assistant that can handle a range of tasks across different Microsoft applications.
The company’s ability to offer OpenAI’s technology without a direct subscription cost could make Copilot a more attractive option for users who might otherwise consider OpenAI’s paid ChatGPT plans.
Think Deeper/o1 vs. DeepSeek R1 vs. Qwen 2.5-Max
DeepSeek, another emerging player in the AI space, has gained traction with its DeepSeek R1 model, an open-source AI system focused on structured reasoning. While the model has shown strong performance in mathematical benchmarks, it has also drawn criticism for accuracy issues.

According to an audit by NewsGuard, DeepSeek’s chatbot failed 83% of accuracy tests, frequently reinforcing false claims instead of debunking them. Microsoft’s decision to make Think Deeper free could increase pressure on alternative models like DeepSeek R1, which has positioned itself as a lower-cost competitor to OpenAI’s proprietary systems.
There is also Alibaba’s Qwen 2.5-Max, which competes with OpenAI’s o1 model and DeepSeek’s R1 reasoning model by offering a hybrid approach that balances efficiency and reasoning power.
While o1 is optimized for structured multi-step reasoning and remains proprietary to OpenAI’s ecosystem, and DeepSeek R1 focuses on reinforcement learning and mathematical problem-solving, Qwen 2.5-Max integrates Mixture-of-Experts (MoE) techniques to optimize resource usage while improving coding, problem-solving, and knowledge-based reasoning tasks.
Benchmark results suggest Qwen 2.5-Max outperforms DeepSeek V3, DeepSeek’s other model which was used to develop DeepSeek R1, in Arena-Hard, LiveBench, and GPQA-Diamond, while maintaining OpenAI API compatibility for easy adoption.

Unlike OpenAI’s closed ecosystem, Alibaba provides access to Qwen 2.5-Max via Alibaba Cloud, making it more accessible for developers. However, OpenAI’s o1 Pro Mode and upcoming o3 series still represent its most advanced reasoning capabilities, placing Alibaba’s model between DeepSeek’s efficiency-focused approach and OpenAI’s premium-tier AI.
Microsoft’s Strategy
The expansion of Think Deeper also aligns with Microsoft’s broader push to make AI more accessible while increasing user engagement with its ecosystem. Suleyman framed the move as a way to reduce friction in handling complex problems, writing, “But in the meantime life is full of chaos and complex problems. Let Think Deeper shoulder some of the mental load!”
The decision to open up Think Deeper for free raises questions about OpenAI’s future monetization strategy. While OpenAI benefits financially from its enterprise agreements with Microsoft, the direct-to-consumer side of its business relies on paid subscriptions.
With Microsoft now providing free access to o1 through Copilot, some users may opt to use Microsoft’s tools instead of paying for ChatGPT Plus or Pro, potentially affecting OpenAI’s revenue from individual subscribers.
Microsoft’s move signals a broader shift in how AI-powered reasoning models are distributed. By removing the paywall for o1, the company is positioning Copilot as a leading AI assistant, directly competing with OpenAI’s ChatGPT, Google’s Gemini, and open-source alternatives like DeepSeek.
Whether OpenAI will adjust its pricing strategy in response remains to be seen, but Microsoft’s decision underscores its ambition to integrate AI seamlessly into its software products while making advanced reasoning models more widely accessible.
Last Updated on February 2, 2025 10:54 am CET