OpenAI has slashed prices for its GPT-5.6 Luna and Terra artificial intelligence models, reducing the cost of the smaller Luna model by 80% and the mid-tier Terra model by 20%. The pricing adjustment, which leaves the flagship Sol model unchanged, responds to growing enterprise caution regarding rising AI expenses and intensifying industry competition. By leveraging efficiency improvements that allow smaller systems to handle tasks previously requiring higher-capability models, OpenAI aims to broaden adoption while navigating a market where businesses are shifting from flat subscription plans to unpredictable usage-based billing.

Pricing Structure

The strategic reset distinguishes between model variants based on their respective cost advantages, targeting different segments of the developer and enterprise market.

Competitive Landscape

The price cuts highlight the intensifying battle for AI market share. OpenAI faces direct competition from Anthropic’s Claude models, which have gained significant traction among enterprise customers, as well as lower-cost open-source models from Chinese AI developers challenging U.S. leaders on both price and performance. This competitive pressure forces OpenAI to balance accessibility with profitability in a rapidly evolving sector.

What the Numbers Show

While cheaper models could drive broader adoption and increase usage volumes, analysts warn the strategy could pressure margins for AI companies investing billions into infrastructure and computing capacity. The divergence between the aggressive 80% cut for the Luna model and the modest 20% reduction for Terra suggests a tiered approach: maximizing volume in high-sensitivity applications while preserving value in mid-tier use cases. This dynamic underscores the tension between scaling user base and maintaining profitability amidst heavy capital expenditure requirements.

OpenAI has rehired Lilian Weng, co-founder of Thinking Machines, to lead a new internal research team focused on accelerating AI capabilities through recursive self-improvement. Weng’s return follows her recent departure from the startup, where she cited health concerns and the unsustainable pace of startup life as primary reasons for stepping down. This move underscores OpenAI’s continued strategy of attracting top-tier talent from competitor firms to bolster its research and development efforts.

Weng announced her departure from Thinking Machines in a post on X, stating that the consistent stress and workload had pushed her beyond what her health could sustain physically. "I don’t feel I’m able to continue at the pace a startup requires," she wrote. Shortly after, OpenAI confirmed her rehire to TechCrunch, noting that she will spearhead research into recursive self-improvement—a process where AI systems enhance their own capabilities through repeated iterations.

Research Focus and Background

The new team under Weng’s leadership will concentrate on advancing OpenAI’s internal AI research. Recursive self-improvement represents a critical area in artificial intelligence development, aiming to create systems that can autonomously refine their performance. Weng previously spent more than six years at OpenAI, holding several key roles including research scientist, head of applied AI research, head of safety systems, and finally VP of research and safety until 2024. She left for Thinking Labs, which later became Thinking Machines.

Thinking Machines co-founder and former OpenAI CTO Mira Murati responded to Weng’s announcement on X, expressing support for her decision to prioritize health. "We’ll miss you, it’s been wonderful building Thinky together. I’m glad that you’re putting you’re health first," Murati wrote. It remains unclear whether Murati was aware of Weng’s imminent return to OpenAI at the time of her comment.

Broader Hiring Trends

Weng’s return is part of a larger trend of OpenAI recruiting high-profile talent from rival companies. Last month, Liz Wamai joined as head of recruiting after three years at Netflix, while Prabhjeet Singh, former Uber Technologies India and South Asia President, was appointed managing director for the country. Additionally, Dean Ball, White House artificial intelligence adviser, joined OpenAI to shape frontier AI policy.

OpenAI has also added 40 employees from Salesforce since the start of the year, as part of a plan to double its workforce from 4,500 to 8,000 by the end of 2026, according to Financial Times reports. Other recent hires include Jason Boehmig, CEO of Ironclad, who will lead the product team for the legal industry, and Brian Landsman, former Salesforce AgentExchange CEO, appointed vice president of global partnerships. Denise Dresser, who led Salesforce’s Slack business, joined as chief revenue officer in December.

What the Numbers Show

The aggressive hiring strategy indicates OpenAI’s intent to scale its operational and research capabilities significantly. Doubling its workforce to 8,000 employees by late 2026 suggests a major expansion in both technical and commercial functions. The recruitment of executives from diverse sectors—technology, entertainment, and government—highlights a multidisciplinary approach to building its next-generation AI infrastructure.