Perceptron Mk1 Video AI Model Undercuts OpenAI by 80-90%

Perceptron Inc. launched its flagship Mk1 video analysis AI model on Monday, pricing it at $0.15 per million input tokens and $1.50 per million output tokens — roughly 80-90% cheaper than competing models from Anthropic, OpenAI, and Google. The two-year-old startup’s model targets enterprise video analysis applications including security monitoring, content editing, and behavioral analysis.

According to Perceptron’s announcement, the company spent 16 months developing what CEO Armen Aghajanyan calls a “multi-modal recipe” designed to understand physical world dynamics. Aghajanyan, formerly of Meta FAIR and Microsoft, positioned the model as addressing enterprise needs for real-time video understanding at scale.

Pricing Strategy Targets Enterprise Adoption

Perceptron’s aggressive pricing strategy directly challenges established players in the video AI market. While Anthropic’s Claude Sonnet 4.5, OpenAI’s GPT-5, and Google’s Gemini 3.1 Pro command premium rates for video analysis, Mk1’s sub-$2 output pricing could accelerate enterprise adoption across security, marketing, and research applications.

The model supports multiple enterprise use cases including automated video editing for social media content, quality control for marketing materials, and behavioral analysis for hiring processes. A public demo allows potential customers to test the model’s capabilities before committing to API integration.

Enterprise video AI represents a growing market as organizations seek to automate visual content analysis. Perceptron’s pricing approach mirrors successful SaaS strategies where lower costs drive higher volume adoption, particularly among mid-market companies previously priced out of advanced AI capabilities.

Thinking Machines Previews Real-Time AI Interaction

Separately, Thinking Machines announced a research preview of what it terms “interaction models” — AI systems designed for real-time, fluid conversation rather than traditional turn-based chat. The startup, founded by former OpenAI CTO Mira Murati and researcher John Schulman, positions these models as addressing limitations in current AI interaction paradigms.

According to Thinking Machines’ blog post, the new architecture treats interactivity as a core model component rather than an external software layer. This approach reportedly reduces latency while enabling more natural voice and video conversations.

The company plans a limited research preview in coming months, though no public availability timeline has been announced. The development represents a shift from current AI models that require users to wait for complete responses before providing additional input.

Apple Ships iOS 26.5 with Encrypted RCS Messaging

Apple released iOS 26.5 for iPhone 11 and newer models, introducing end-to-end encrypted RCS messaging between iPhones and Android devices. The update, currently in beta and requiring carrier support, addresses long-standing interoperability issues between iOS and Android messaging systems.

The release notably excludes the anticipated new Siri features, with analysts now expecting that debut at WWDC on June 8 with iOS 27. Additional features include a “Pride Luminance” wallpaper and “Suggested Places” in Maps based on local trends and user activity.

Google’s Android ecosystem president Sameer Samat welcomed the change on X, calling it “big news” for cross-platform messaging security. The RCS implementation represents Apple’s response to regulatory pressure for improved Android compatibility.

Legal Battle Continues Over OpenAI’s Direction

The high-stakes trial between Elon Musk and OpenAI entered its third week, with closing arguments scheduled for May 14. Musk’s 2024 lawsuit alleges OpenAI abandoned its founding mission of developing AI for humanity’s benefit in favor of profit maximization.

Microsoft CEO Satya Nadella and former OpenAI chief scientist Ilya Sutskever testified this week, following earlier testimony from OpenAI cofounder Greg Brockman and former CTO Mira Murati. Sam Altman took the stand Tuesday to counter Musk’s characterizations of his leadership.

OpenAI maintains that Musk’s claims lack merit, stating the lawsuit represents an attempt to gain control over the company’s direction. The trial outcome could significantly impact OpenAI’s governance structure and future development priorities.

What This Means

The convergence of aggressive pricing from Perceptron, real-time interaction research from Thinking Machines, and ongoing governance battles at OpenAI reflects an AI industry in rapid transition. Perceptron’s 80-90% cost reduction strategy could force established players to reconsider their pricing models, particularly as enterprise video AI adoption accelerates.

Thinking Machines’ interaction model research suggests the industry is moving beyond current chat-based paradigms toward more natural, continuous AI conversation. This shift could reshape user expectations and competitive dynamics as real-time AI becomes technically feasible.

The Musk-OpenAI trial highlights fundamental tensions between commercial success and mission-driven AI development. The outcome may establish precedents for how AI companies balance profit motives with stated public benefit goals, influencing future governance structures across the industry.

FAQ

How does Perceptron Mk1’s pricing compare to competitors?

Perceptron Mk1 costs $0.15 per million input tokens and $1.50 per million output tokens, which is 80-90% cheaper than similar video analysis capabilities from Anthropic’s Claude Sonnet 4.5, OpenAI’s GPT-5, and Google’s Gemini 3.1 Pro.

What makes Thinking Machines’ interaction models different?

Unlike current AI models that operate on turn-based chat, Thinking Machines’ interaction models are designed for real-time, fluid conversation where the AI can process and respond to inputs continuously rather than waiting for complete user queries.

When will the Musk vs OpenAI trial conclude?

Closing arguments in the Musk vs OpenAI trial are scheduled for May 14, 2026. The trial centers on Musk’s claims that OpenAI abandoned its founding mission of developing AI for humanity’s benefit in favor of profit maximization.