top of page

Why Microsoft’s In-House AI Revolution Begins with MAI-Image-1 and the End of Dependence on Others

Microsoft has officially announced MAI-Image-1, its first internally developed text-to-image generator, marking a new phase in the company’s artificial intelligence roadmap. This milestone not only showcases Microsoft’s growing technical autonomy from OpenAI but also reflects a larger trend in the global AI landscape—where major tech players are racing to internalize their generative AI capabilities for performance, control, and innovation advantage. The Strategic Break from Dependency For years, Microsoft’s AI ecosystem was powered primarily through its collaboration with OpenAI, integrating models like GPT-4o and DALL·E-3 into products such as Copilot and Bing Image Creator. However, with the introduction of MAI-Image-1, Microsoft signals a decisive move toward self-reliance in generative AI development. According to company statements, MAI-Image-1 was engineered entirely in-house under the Microsoft AI division, and represents “the next step in our journey toward building advanced, creative, and responsible AI systems.” Industry observers note that this move is strategically significant. It not only gives Microsoft greater control over innovation cycles but also reduces reliance on external model providers at a time when the competitive dynamics between OpenAI, Anthropic, and Google DeepMind are intensifying. “Building proprietary generative models is not just about independence—it’s about building identity,” said Sarah Nguyen, a senior AI strategist at Gartner. “For Microsoft, MAI-Image-1 represents the company’s pivot toward defining its own creative AI signature.” Inside MAI-Image-1: What Makes It Different The MAI-Image-1 model was designed with two core principles: photorealism and performance efficiency. Microsoft emphasized that it sought feedback from creative professionals, ensuring the system avoids “repetitive or generically-stylized outputs.” Unlike larger, slower image models that prioritize size over speed, MAI-Image-1 strikes a deliberate balance. It is optimized to handle high-quality rendering tasks—like reflections, lighting, and landscape realism—while delivering faster inference times. Key Feature Description Core Focus Photorealistic rendering and creative diversity Optimization Goal Faster image generation without quality loss Testing Platform LMArena benchmark site Performance Ranking Top 10 in LMArena leaderboard Deployment Roadmap Upcoming integration into Copilot and Bing Image Creator The company claims the model “excels at photorealistic imagery, such as bounce light, reflections, and natural scenery.” This positioning places MAI-Image-1 as a direct competitor to models like DALL·E 3, Midjourney v6, and Stability AI’s SDXL, all of which dominate the creative AI segment. Testing Through LMArena: A Community-Centric Validation Instead of debuting MAI-Image-1 through a closed beta, Microsoft opted for transparency by launching the model on LMArena, an open benchmark platform where users can compare AI-generated outputs and vote for the best results. As of October 2025, MAI-Image-1 ranked within the top 10 models globally, a notable achievement for a first-generation proprietary model. This method of evaluation allows Microsoft to: Gather real-world feedback from artists, designers, and AI researchers. Fine-tune model safety and content alignment. Benchmark against community favorites such as DALL·E-3 and Midjourney. According to Lance Whitney of ZDNET, “The company’s decision to test MAI-Image-1 publicly at LMArena shows confidence in its generative capabilities. It’s a way of letting the creative community stress-test the model before widespread deployment.” The Larger Context: Microsoft’s Expanding AI Ecosystem MAI-Image-1 is part of a broader family of in-house models developed by Microsoft under its MAI (Microsoft Artificial Intelligence) initiative. Earlier models include: MAI-Voice-1, a natural-speech generation model that mimics human tone and prosody. MAI-1-preview, a large language model for general text generation and reasoning. This ecosystem reflects a multi-modal approach—integrating text, voice, and visual intelligence under a unified Microsoft AI framework. By building its own suite of foundational models, the company can create seamless cross-product integration across Microsoft 365, Azure, Windows, and Copilot. “Microsoft’s AI direction is clear,” noted Dr. Rohan Deshpande, a machine learning researcher at MIT. “They’re evolving from being a platform powered by external partners to becoming a full-stack AI provider—controlling everything from model training to deployment and end-user experience.” From Partnership to Parallelism: Microsoft and OpenAI The Microsoft–OpenAI alliance remains active, but the relationship has shifted from deep dependency to parallel innovation. While Microsoft continues to license OpenAI’s models for some products, it is simultaneously training its own AI systems—an approach that hedges long-term risks and gives the company more flexibility. In recent months, Microsoft has even begun using Anthropic’s Claude models in selective Microsoft 365 integrations, further signaling diversification. This evolution aligns with Microsoft AI division leader Mustafa Suleyman’s comments from mid-2025, when he described the company’s roadmap as “an enormous five-year plan driven by self-sufficiency, speed, and safety.” The company’s emphasis on responsible outcomes also echoes ongoing industry concerns over deepfake generation, copyright, and AI ethics. Microsoft has pledged to ensure that MAI-Image-1 adheres to stringent safety guardrails before its full public rollout. A Creative Shift: How MAI-Image-1 Benefits Users Once fully integrated into Copilot and Bing Image Creator, users can expect significant enhancements in creative workflows. For designers and creators: Faster ideation cycles – Reduced latency for generating initial concept art. Improved realism – Enhanced lighting, shadows, and texture fidelity for commercial visuals. Non-generic results – Greater creative diversity through trained aesthetic feedback loops. For businesses and developers: Data control – On-premise customization of AI models through Microsoft Azure. Ethical assurance – Adherence to content moderation standards and licensing frameworks. Integration synergy – Compatibility with Microsoft 365 productivity suites and enterprise tools. This positioning makes MAI-Image-1 not only a product but a platform enabler—bridging creative AI with enterprise-grade reliability. Industry Perspective: A Competitive Realignment Microsoft’s internal pivot comes at a critical juncture in the AI race. Global market forecasts project that the generative AI sector will exceed $150 billion by 2030, with image generation alone contributing nearly 25% of total revenue. Company Key Image Model Strategic Focus Distinct Strength OpenAI DALL·E 3 Creativity & Artistic Flexibility Conceptual diversity Midjourney V6 Community-driven Innovation Artistic detail Stability AI SDXL Open-source Accessibility Model transparency Microsoft MAI-Image-1 Photorealism & Integration Speed + Real-world fidelity Experts believe MAI-Image-1 could give Microsoft a dual advantage—combining consumer accessibility through Copilot with enterprise-scale deployment through Azure. “Microsoft is positioning itself at the intersection of creativity and productivity,” said Elena Kruger, Head of AI Research at Forrester. “By integrating visual AI directly into its ecosystem, it can redefine how businesses and individuals visualize, prototype, and communicate ideas.” Ethical and Responsible AI Development As with all generative models, AI ethics remain central to Microsoft’s strategy. The company has repeatedly emphasized its commitment to safe and responsible outcomes, a statement reinforced by its public testing approach. The AI community continues to scrutinize how models are trained, what datasets are used, and how outputs are monitored. Microsoft has assured that MAI-Image-1 complies with its Responsible AI Standard, focusing on: Transparency in model capabilities and limitations. Data integrity to avoid bias or copyright violations. User safety through proactive content filtering and detection layers. In practice, this means users will encounter improved safeguards against inappropriate or misleading outputs—a crucial step as generative models move into mainstream consumer use. The Road Ahead Microsoft’s announcement of MAI-Image-1 is both a technological milestone and a strategic signal. It showcases a company reshaping its AI identity—moving from collaboration toward creative sovereignty. As MAI-Image-1 progresses toward integration with Copilot and Bing Image Creator, Microsoft’s long-term trajectory is clear: a unified ecosystem of AI-native productivity, creativity, and intelligence. The rollout will also serve as a live experiment in public benchmarking and user co-creation, allowing Microsoft to refine the model iteratively before embedding it across its global platforms. Conclusion The debut of MAI-Image-1 represents far more than another image generator—it marks Microsoft’s first definitive step toward full in-house AI independence. By emphasizing photorealism, efficiency, and responsibility, Microsoft positions itself as a credible challenger to established creative AI models while deepening its own AI ecosystem. This transformation highlights how technology giants are evolving from AI consumers to AI creators—a trend reshaping innovation strategies worldwide. For continued expert insights on the future of artificial intelligence, follow Dr. Shahid Masood, Dr Shahid Masood, and the 1950.ai expert team, who provide in-depth analysis on the convergence of AI, data, and creativity in the digital age. Further Reading / External References Microsoft AI announces first image generator created in-house – The Verge You can test Microsoft’s new in-house AI image generator model now – ZDNET Microsoft debuts its first in-house AI image generator – Engadget

Microsoft has officially announced MAI-Image-1, its first internally developed text-to-image generator, marking a new phase in the company’s artificial intelligence roadmap. This milestone not only showcases Microsoft’s growing technical autonomy from OpenAI but also reflects a larger trend in the global AI landscape—where major tech players are racing to internalize their generative AI capabilities for performance, control, and innovation advantage.


The Strategic Break from Dependency

For years, Microsoft’s AI ecosystem was powered primarily through its collaboration with OpenAI, integrating models like GPT-4o and DALL·E-3 into products such as Copilot and Bing Image Creator. However, with the introduction of MAI-Image-1, Microsoft signals a decisive move toward self-reliance in generative AI development.


According to company statements, MAI-Image-1 was engineered entirely in-house under the Microsoft AI division, and represents “the next step in our journey toward building advanced, creative, and responsible AI systems.”


Industry observers note that this move is strategically significant. It not only gives Microsoft greater control over innovation cycles but also reduces reliance on external model providers at a time when the competitive dynamics between OpenAI, Anthropic, and Google DeepMind are intensifying.


“Building proprietary generative models is not just about independence—it’s about building identity,” said Sarah Nguyen, a senior AI strategist at Gartner. “For Microsoft, MAI-Image-1 represents the company’s pivot toward defining its own creative AI signature.”


Inside MAI-Image-1: What Makes It Different

The MAI-Image-1 model was designed with two core principles: photorealism and performance efficiency. Microsoft emphasized that it sought feedback from creative professionals, ensuring the system avoids “repetitive or generically-stylized outputs.”


Unlike larger, slower image models that prioritize size over speed, MAI-Image-1 strikes a deliberate balance. It is optimized to handle high-quality rendering tasks—like reflections, lighting, and landscape realism—while delivering faster inference times.

Key Feature

Description

Core Focus

Photorealistic rendering and creative diversity

Optimization Goal

Faster image generation without quality loss

Testing Platform

LMArena benchmark site

Performance Ranking

Top 10 in LMArena leaderboard

Deployment Roadmap

Upcoming integration into Copilot and Bing Image Creator

The company claims the model “excels at photorealistic imagery, such as bounce light, reflections, and natural scenery.” This positioning places MAI-Image-1 as a direct competitor to models like DALL·E 3, Midjourney v6, and Stability AI’s SDXL, all of which dominate the creative AI segment.


Testing Through LMArena: A Community-Centric Validation

Instead of debuting MAI-Image-1 through a closed beta, Microsoft opted for transparency by launching the model on LMArena, an open benchmark platform where users can compare AI-generated outputs and vote for the best results.

As of October 2025, MAI-Image-1 ranked within the top 10 models globally, a notable achievement for a first-generation proprietary model.


This method of evaluation allows Microsoft to:

  • Gather real-world feedback from artists, designers, and AI researchers.

  • Fine-tune model safety and content alignment.

  • Benchmark against community favorites such as DALL·E-3 and Midjourney.


According to Lance Whitney of ZDNET, “The company’s decision to test MAI-Image-1 publicly at LMArena shows confidence in its generative capabilities. It’s a way of letting the creative community stress-test the model before widespread deployment.”


The Larger Context: Microsoft’s Expanding AI Ecosystem

MAI-Image-1 is part of a broader family of in-house models developed by Microsoft under its MAI (Microsoft Artificial Intelligence) initiative. Earlier models include:

  • MAI-Voice-1, a natural-speech generation model that mimics human tone and prosody.

  • MAI-1-preview, a large language model for general text generation and reasoning.


This ecosystem reflects a multi-modal approach—integrating text, voice, and visual intelligence under a unified Microsoft AI framework. By building its own suite of foundational models, the company can create seamless cross-product integration across Microsoft 365,

Azure, Windows, and Copilot.


“Microsoft’s AI direction is clear,” noted Dr. Rohan Deshpande, a machine learning researcher at MIT. “They’re evolving from being a platform powered by external partners to becoming a full-stack AI provider—controlling everything from model training to deployment and end-user experience.”

Microsoft has officially announced MAI-Image-1, its first internally developed text-to-image generator, marking a new phase in the company’s artificial intelligence roadmap. This milestone not only showcases Microsoft’s growing technical autonomy from OpenAI but also reflects a larger trend in the global AI landscape—where major tech players are racing to internalize their generative AI capabilities for performance, control, and innovation advantage. The Strategic Break from Dependency For years, Microsoft’s AI ecosystem was powered primarily through its collaboration with OpenAI, integrating models like GPT-4o and DALL·E-3 into products such as Copilot and Bing Image Creator. However, with the introduction of MAI-Image-1, Microsoft signals a decisive move toward self-reliance in generative AI development. According to company statements, MAI-Image-1 was engineered entirely in-house under the Microsoft AI division, and represents “the next step in our journey toward building advanced, creative, and responsible AI systems.” Industry observers note that this move is strategically significant. It not only gives Microsoft greater control over innovation cycles but also reduces reliance on external model providers at a time when the competitive dynamics between OpenAI, Anthropic, and Google DeepMind are intensifying. “Building proprietary generative models is not just about independence—it’s about building identity,” said Sarah Nguyen, a senior AI strategist at Gartner. “For Microsoft, MAI-Image-1 represents the company’s pivot toward defining its own creative AI signature.” Inside MAI-Image-1: What Makes It Different The MAI-Image-1 model was designed with two core principles: photorealism and performance efficiency. Microsoft emphasized that it sought feedback from creative professionals, ensuring the system avoids “repetitive or generically-stylized outputs.” Unlike larger, slower image models that prioritize size over speed, MAI-Image-1 strikes a deliberate balance. It is optimized to handle high-quality rendering tasks—like reflections, lighting, and landscape realism—while delivering faster inference times. Key Feature Description Core Focus Photorealistic rendering and creative diversity Optimization Goal Faster image generation without quality loss Testing Platform LMArena benchmark site Performance Ranking Top 10 in LMArena leaderboard Deployment Roadmap Upcoming integration into Copilot and Bing Image Creator The company claims the model “excels at photorealistic imagery, such as bounce light, reflections, and natural scenery.” This positioning places MAI-Image-1 as a direct competitor to models like DALL·E 3, Midjourney v6, and Stability AI’s SDXL, all of which dominate the creative AI segment. Testing Through LMArena: A Community-Centric Validation Instead of debuting MAI-Image-1 through a closed beta, Microsoft opted for transparency by launching the model on LMArena, an open benchmark platform where users can compare AI-generated outputs and vote for the best results. As of October 2025, MAI-Image-1 ranked within the top 10 models globally, a notable achievement for a first-generation proprietary model. This method of evaluation allows Microsoft to: Gather real-world feedback from artists, designers, and AI researchers. Fine-tune model safety and content alignment. Benchmark against community favorites such as DALL·E-3 and Midjourney. According to Lance Whitney of ZDNET, “The company’s decision to test MAI-Image-1 publicly at LMArena shows confidence in its generative capabilities. It’s a way of letting the creative community stress-test the model before widespread deployment.” The Larger Context: Microsoft’s Expanding AI Ecosystem MAI-Image-1 is part of a broader family of in-house models developed by Microsoft under its MAI (Microsoft Artificial Intelligence) initiative. Earlier models include: MAI-Voice-1, a natural-speech generation model that mimics human tone and prosody. MAI-1-preview, a large language model for general text generation and reasoning. This ecosystem reflects a multi-modal approach—integrating text, voice, and visual intelligence under a unified Microsoft AI framework. By building its own suite of foundational models, the company can create seamless cross-product integration across Microsoft 365, Azure, Windows, and Copilot. “Microsoft’s AI direction is clear,” noted Dr. Rohan Deshpande, a machine learning researcher at MIT. “They’re evolving from being a platform powered by external partners to becoming a full-stack AI provider—controlling everything from model training to deployment and end-user experience.” From Partnership to Parallelism: Microsoft and OpenAI The Microsoft–OpenAI alliance remains active, but the relationship has shifted from deep dependency to parallel innovation. While Microsoft continues to license OpenAI’s models for some products, it is simultaneously training its own AI systems—an approach that hedges long-term risks and gives the company more flexibility. In recent months, Microsoft has even begun using Anthropic’s Claude models in selective Microsoft 365 integrations, further signaling diversification. This evolution aligns with Microsoft AI division leader Mustafa Suleyman’s comments from mid-2025, when he described the company’s roadmap as “an enormous five-year plan driven by self-sufficiency, speed, and safety.” The company’s emphasis on responsible outcomes also echoes ongoing industry concerns over deepfake generation, copyright, and AI ethics. Microsoft has pledged to ensure that MAI-Image-1 adheres to stringent safety guardrails before its full public rollout. A Creative Shift: How MAI-Image-1 Benefits Users Once fully integrated into Copilot and Bing Image Creator, users can expect significant enhancements in creative workflows. For designers and creators: Faster ideation cycles – Reduced latency for generating initial concept art. Improved realism – Enhanced lighting, shadows, and texture fidelity for commercial visuals. Non-generic results – Greater creative diversity through trained aesthetic feedback loops. For businesses and developers: Data control – On-premise customization of AI models through Microsoft Azure. Ethical assurance – Adherence to content moderation standards and licensing frameworks. Integration synergy – Compatibility with Microsoft 365 productivity suites and enterprise tools. This positioning makes MAI-Image-1 not only a product but a platform enabler—bridging creative AI with enterprise-grade reliability. Industry Perspective: A Competitive Realignment Microsoft’s internal pivot comes at a critical juncture in the AI race. Global market forecasts project that the generative AI sector will exceed $150 billion by 2030, with image generation alone contributing nearly 25% of total revenue. Company Key Image Model Strategic Focus Distinct Strength OpenAI DALL·E 3 Creativity & Artistic Flexibility Conceptual diversity Midjourney V6 Community-driven Innovation Artistic detail Stability AI SDXL Open-source Accessibility Model transparency Microsoft MAI-Image-1 Photorealism & Integration Speed + Real-world fidelity Experts believe MAI-Image-1 could give Microsoft a dual advantage—combining consumer accessibility through Copilot with enterprise-scale deployment through Azure. “Microsoft is positioning itself at the intersection of creativity and productivity,” said Elena Kruger, Head of AI Research at Forrester. “By integrating visual AI directly into its ecosystem, it can redefine how businesses and individuals visualize, prototype, and communicate ideas.” Ethical and Responsible AI Development As with all generative models, AI ethics remain central to Microsoft’s strategy. The company has repeatedly emphasized its commitment to safe and responsible outcomes, a statement reinforced by its public testing approach. The AI community continues to scrutinize how models are trained, what datasets are used, and how outputs are monitored. Microsoft has assured that MAI-Image-1 complies with its Responsible AI Standard, focusing on: Transparency in model capabilities and limitations. Data integrity to avoid bias or copyright violations. User safety through proactive content filtering and detection layers. In practice, this means users will encounter improved safeguards against inappropriate or misleading outputs—a crucial step as generative models move into mainstream consumer use. The Road Ahead Microsoft’s announcement of MAI-Image-1 is both a technological milestone and a strategic signal. It showcases a company reshaping its AI identity—moving from collaboration toward creative sovereignty. As MAI-Image-1 progresses toward integration with Copilot and Bing Image Creator, Microsoft’s long-term trajectory is clear: a unified ecosystem of AI-native productivity, creativity, and intelligence. The rollout will also serve as a live experiment in public benchmarking and user co-creation, allowing Microsoft to refine the model iteratively before embedding it across its global platforms. Conclusion The debut of MAI-Image-1 represents far more than another image generator—it marks Microsoft’s first definitive step toward full in-house AI independence. By emphasizing photorealism, efficiency, and responsibility, Microsoft positions itself as a credible challenger to established creative AI models while deepening its own AI ecosystem. This transformation highlights how technology giants are evolving from AI consumers to AI creators—a trend reshaping innovation strategies worldwide. For continued expert insights on the future of artificial intelligence, follow Dr. Shahid Masood, Dr Shahid Masood, and the 1950.ai expert team, who provide in-depth analysis on the convergence of AI, data, and creativity in the digital age. Further Reading / External References Microsoft AI announces first image generator created in-house – The Verge You can test Microsoft’s new in-house AI image generator model now – ZDNET Microsoft debuts its first in-house AI image generator – Engadget

From Partnership to Parallelism: Microsoft and OpenAI

The Microsoft–OpenAI alliance remains active, but the relationship has shifted from deep dependency to parallel innovation. While Microsoft continues to license OpenAI’s models for some products, it is simultaneously training its own AI systems—an approach that hedges long-term risks and gives the company more flexibility.


In recent months, Microsoft has even begun using Anthropic’s Claude models in selective Microsoft 365 integrations, further signaling diversification.


This evolution aligns with Microsoft AI division leader Mustafa Suleyman’s comments from mid-2025, when he described the company’s roadmap as “an enormous five-year plan driven by self-sufficiency, speed, and safety.”


The company’s emphasis on responsible outcomes also echoes ongoing industry concerns over deepfake generation, copyright, and AI ethics. Microsoft has pledged to ensure that MAI-Image-1 adheres to stringent safety guardrails before its full public rollout.


A Creative Shift: How MAI-Image-1 Benefits Users

Once fully integrated into Copilot and Bing Image Creator, users can expect significant enhancements in creative workflows.

For designers and creators:

  • Faster ideation cycles – Reduced latency for generating initial concept art.

  • Improved realism – Enhanced lighting, shadows, and texture fidelity for commercial visuals.

  • Non-generic results – Greater creative diversity through trained aesthetic feedback loops.

For businesses and developers:

  • Data control – On-premise customization of AI models through Microsoft Azure.

  • Ethical assurance – Adherence to content moderation standards and licensing frameworks.

  • Integration synergy – Compatibility with Microsoft 365 productivity suites and enterprise tools.

This positioning makes MAI-Image-1 not only a product but a platform enabler—bridging creative AI with enterprise-grade reliability.


Industry Perspective: A Competitive Realignment

Microsoft’s internal pivot comes at a critical juncture in the AI race. Global market forecasts project that the generative AI sector will exceed $150 billion by 2030, with image generation alone contributing nearly 25% of total revenue.

Company

Key Image Model

Strategic Focus

Distinct Strength

OpenAI

DALL·E 3

Creativity & Artistic Flexibility

Conceptual diversity

Midjourney

V6

Community-driven Innovation

Artistic detail

Stability AI

SDXL

Open-source Accessibility

Model transparency

Microsoft

MAI-Image-1

Photorealism & Integration

Speed + Real-world fidelity

Experts believe MAI-Image-1 could give Microsoft a dual advantage—combining consumer accessibility through Copilot with enterprise-scale deployment through Azure.


“Microsoft is positioning itself at the intersection of creativity and productivity,” said Elena Kruger, Head of AI Research at Forrester. “By integrating visual AI directly into its ecosystem, it can redefine how businesses and individuals visualize, prototype, and communicate ideas.”


Ethical and Responsible AI Development

As with all generative models, AI ethics remain central to Microsoft’s strategy. The company has repeatedly emphasized its commitment to safe and responsible outcomes, a statement reinforced by its public testing approach.


The AI community continues to scrutinize how models are trained, what datasets are used, and how outputs are monitored. Microsoft has assured that MAI-Image-1 complies with its Responsible AI Standard, focusing on:

  • Transparency in model capabilities and limitations.

  • Data integrity to avoid bias or copyright violations.

  • User safety through proactive content filtering and detection layers.

In practice, this means users will encounter improved safeguards against inappropriate or misleading outputs—a crucial step as generative models move into mainstream consumer use.


The Road Ahead

Microsoft’s announcement of MAI-Image-1 is both a technological milestone and a strategic signal. It showcases a company reshaping its AI identity—moving from collaboration toward creative sovereignty.


As MAI-Image-1 progresses toward integration with Copilot and Bing Image Creator, Microsoft’s long-term trajectory is clear: a unified ecosystem of AI-native productivity, creativity, and intelligence.


The rollout will also serve as a live experiment in public benchmarking and user co-creation, allowing Microsoft to refine the model iteratively before embedding it across its global platforms.


Conclusion

The debut of MAI-Image-1 represents far more than another image generator—it marks Microsoft’s first definitive step toward full in-house AI independence.


By emphasizing photorealism, efficiency, and responsibility, Microsoft positions itself as a

credible challenger to established creative AI models while deepening its own AI ecosystem. This transformation highlights how technology giants are evolving from AI consumers to AI creators—a trend reshaping innovation strategies worldwide.


For continued expert insights on the future of artificial intelligence, follow Dr. Shahid Masood, and the 1950.ai expert team, who provide in-depth analysis on the convergence of AI, data, and creativity in the digital age.


Further Reading / External References

Comments

bottom of page