Microsoft Corporation (MSFT.US) Unveils Six Specialized AI Models at Ignite Conference: Challenging OpenAI and Anthropic with Low Costs

date
07:24 18/11/2026
avatar
GMT Eight
Microsoft is making significant progress in building its own frontier general-purpose AI models, while unveiling six specialized AI models at the Ignite conference.
Microsoft Corporation (MSFT.US) is accelerating its in-house AI model development. Mustafa Suleyman, Microsoft's AI chief, said the company is making significant progress in building its own frontier general-purpose AI models, while unveiling six specialized AI models at the Ignite conference and claiming they can achieve comparable performance at a fraction of the cost of leading products from OpenAI and Anthropic. Microsoft said some tasks that previously took weeks can now be completed in minutes. Key Highlights: From Reliance on External Models to In-House Competition For a long time, Microsoft's most advanced models have relied on OpenAI and other AI companies. This launch demonstrates that Microsoft can develop competitive proprietary technology and signals its plan to compete aggressively on price. Microsoft unveiled six new AI models at the Ignite conference in San Francisco on Tuesday, covering coding, cybersecurity, image generation and voice. Specifically: MAI Code 2: Microsoft said its coding model achieves 98% of GPT-6.1 Sol's performance at 40% of the latter's price; and 96% of Anthropic Claude Opus 5.5's performance at just 5% of the latter's cost. The flagship coding model is designed to handle complex, time-consuming programming tasks, working across multiple files and understanding entire software projects. MAI Code 2 Flash: A smaller, faster model. Microsoft said its performance matches GPT-6 Luna at 65% of the latter's price; and matches Claude Haiku 5.5 at 20% of the latter's cost. MAI Cyber 2 Flash: A cybersecurity model. Microsoft said it beats GPT-6 Astra on a certain security benchmark at less than 2% of the latter's cost per task. Microsoft also plans to offer cyber-specialized models to a limited customer group to help address the surge in software vulnerabilities increasingly exposed by AI tools. Other models: According to Microsoft, MAI Image 2.7 ranks first in image editing; new transcription and real-time voice models are designed to challenge leading rivals. Strategy: Specialized Models for Cost Advantage Cost pressure on enterprise customers is an important backdrop to this launch. Suleyman said he constantly hears enterprises complain that "token spending is soaring." He said: "Our promise is to offer the best price-performance in the market so you can truly scale agents; and we also know you want models that are truly your own." Microsoft's strategy is to build specialized models targeting specific tasks, thereby achieving comparable performance at significantly lower cost. Suleyman said: "This basically means you pack more tokens, knowledge and expertise into the same number of parameters." He also said: "This is not just a model, not a flash in the pan. This is a system we've built, a machine that keeps climbing, jointly optimizing performance and price." But this strategy involves trade-offs: a model optimized for everyday coding tasks won't be as strong in advanced mathematics or scientific research. In addition, Microsoft's performance comparisons rely on selected benchmarks, and some price advantages also reflect promotional discounts. Enterprise Platform and Cost Control Beyond the models themselves, Microsoft also emphasized enterprise-grade platforms and cost-control tools at Ignite. Executives highlighted the Microsoft IQ platform, designed to help enterprises organize and share employee knowledge, expertise and tools, enabling AI systems to learn how the enterprise operates and continuously build on that foundation. Microsoft CEO Satya Nadella emphasized that the new AI models cover scenarios including coding, cybersecurity and decision-making. He also highlighted Copilot Fusion, saying it can orchestrate multiple models to deliver frontier-level results at significantly lower cost. Microsoft also demonstrated a model-selection feature that can choose the right model for a task, maximizing output quality while managing the cost of each request. Industry Context: Frontier AI Competition Heats Up in Tandem While Microsoft targets industry leaders, frontier AI itself is also evolving rapidly. OpenAI released GPT-6.2 Astra and GPT-6.2 Sol on Tuesday, with the new Astra priced 20% lower than the previous version. Microsoft is also advancing a frontier general-purpose large language model, aiming to compete across a broader range of tasks. Suleyman said Microsoft has achieved state-of-the-art pre-training results and is making strong progress in reinforcement learning, but declined to give a release timeline for the model. He said: "My team is building true superintelligence." Potential Impact: Microsoft's Relationship with OpenAI May Be Reshaped The new models could reshape Microsoft's relationships with OpenAI and other external AI providers. MAI Code 2 will be available in GitHub Copilot, potentially shifting some coding workloads from external models to Microsoft's own models. When asked how much Copilot traffic Microsoft expects to shift to its own models, Suleyman declined to give a target but predicted adoption would be significant. On the enterprise side, Microsoft also offers licensed model weights to enterprise customers, allowing them to customize its models with proprietary data, but requiring deployment within Azure. Suleyman said this approach "feels like open source" while offering greater security and control. Safety and Alignment Listed as Priorities Suleyman said recent cybersecurity incidentsciting the example of OpenAI's AI agent breaching the Hugging Face platformprompted him to focus not only on performance but also on critical safeguards. Preventing AI from taking harmful actions and ensuring it behaves as intended, known as "alignment" research, is his top priority. At the opening of the conference, Nadella also acknowledged the cross-regional service outage that occurred last week, saying the company takes customer trust seriously, has work to do, and pledged continuous learning and improvement. The market will now focus on two things: first, how quickly Microsoft can translate progress in specialized AI into competitive general-purpose models; second, whether its cost advantages are sufficient to prompt customers to switch from established competitors to Microsoft. Microsoft closed down 2.27% on Tuesday, up 5.3% year to date, underperforming the S&P 500. As of press time, the stock was up 0.9% in after-hours trading.