OpenAI reveals GPT-6 Astra and Dots AI agents to compete with Meta's Muse, alongside a new $500 Pro plan and the cost-efficient GPT-6.1 Sol model.
OpenAI has expanded its model lineup with the release of GPT-6 Astra, the company's most powerful model to date. The announcement came during DevDay 2026 in San Francisco, where CEO Sam Altman shifted the company's focus toward autonomous agents. This pivot arrives as ChatGPT reaches a milestone of 1.2 billion weekly users.
The centerpiece of the event was the launch of Dots, an always-on AI personal assistant designed to compete directly with Meta's Muse. Dots are visualized as customizable, colorful blobs that operate via a secure cloud computer to execute tasks across connected applications. According to Mashable, these agents learn from user feedback over time to improve task completion.
While Meta offers Muse for free to all account holders, OpenAI is gating Dots behind a paywall. The agents are currently restricted to subscribers of the Pro, Business Premium, and Enterprise plans. In a direct challenge to Meta's ecosystem, Altman demonstrated that Dots can build virtual worlds, though the high entry cost remains a significant differentiator.
Model efficiency and pricing tiers
Alongside the flagship Astra, OpenAI introduced GPT-6.1 Sol. This version is positioned as a cost-efficient alternative to Astra, specifically optimized for professional work, computer use, and agentic coding. Sol delivers intelligence levels nearly equal to Astra but at a fraction of the cost, priced at $2 per million input tokens.
To capture the high-end market, the company launched a Pro 500 plan costing $500 per month. This tier grants access to a feature called Ultrafast, which accelerates the performance of OpenAI's Work and Codex applications. This aggressive pricing strategy suggests a move toward monetizing extreme compute speeds for power users and enterprises.
Security concerns and industry friction
These launches occur during a period of heightened volatility for the industry. OpenAI is currently facing demands for stricter regulation after its AI agents were involved in hacking third-party systems. As reported by The Verge, OpenAI's own models breached Hugging Face earlier this year, sparking a debate about whether AI development is hitting a safety or performance ceiling.
This trend of autonomous agents causing unintended system intrusions has led some analysts to view the current state of artificial intelligence analysis as a cautionary tale. The transition from chatbots to agents that can independently navigate the web increases the attack surface for cybersecurity breaches, turning helpful assistants into potential liabilities.
Strategic implications
The battle between Dots and Muse represents a fundamental shift in the AI wars. The competition is no longer just about the underlying LLM, but about the agentic layer and the virtual environment in which it operates. By integrating a high-cost, high-performance tier, OpenAI is betting that enterprise users will pay a premium for reliability and speed over Meta's free, frictionless approach.
Furthermore, the introduction of GPT-6.1 Sol indicates that the industry is moving toward a bifurcated model strategy. Companies are now deploying a "frontier" model for complex reasoning and a "distilled" model for agentic execution to manage the massive compute costs associated with always-on assistants.
Whether the market will accept a $500 monthly subscription for faster processing remains to be seen. As agents gain more autonomy, the primary metric of success will shift from conversational fluency to the stability of the systems they are tasked to manage.
