The company says its post-trained models cut AI costs by 10 times while matching frontier-level quality for marketing tasks.
Optimizely, the AI platform for marketing, on Tuesday announced a family of purpose-built, post-trained AI models designed specifically for marketing work. In early testing, the company said, the models achieved 10 times greater cost efficiency than state-of-the-art large language models, or LLMs.
Frontier models are built to be generalists, capable of a wide range of tasks, but that breadth comes at a cost. Marketing tasks are nuanced and domain-specific, so a general model carries more parameters, pulls in extra context and spends tokens it doesn’t need, Optimizely said. The company’s approach strips away that excess from state-of-the-art models, leaving models built specifically for marketing tasks.
Marketers often ask the same general-purpose AI models to handle a variety of jobs, from writing campaign copy to analyzing experiments and interpreting customer behavior. But knowing how to perform a task is different from understanding the brand behind it. In a global study of more than 2,000 B2B marketing leaders, Optimizely found that 53% said current AI tools can capture the facts of a brand but struggle to capture the emotional resonance that helps it connect with audiences. The company said its new models were built around a different premise: that AI can take on more of the work without stripping away the context and understanding that make marketing distinctive.
Also Read: Why AI-Informed Marketing Needs a New Set of Rules
“The initial results of our AI Lab have surpassed our expectations,” said Imran Yousuf, head of AI engineering at Optimizely. “We believe post-training models is the long-term solution to the challenging cost vs. quality debate. Our purpose-built family of models is built to get work done fast and cost-effectively, without sacrificing quality. Customers shouldn’t have to think about which model to use, when, or how. They can just trust that our agent platform will apply the right model for the right use case, so their time and their budget go toward the work that actually grows the business.”
Optimizely is also introducing Mark-Bench, an open-source benchmark built to evaluate AI performance for the marketing domain specifically. The company said its intent is for marketers, researchers and other AI providers to test their own models and agentic harnesses against it, giving the industry a shared, objective way to measure both cost and performance on marketing-specific work. Mark-Bench tests models directly against 285 tasks spanning 15 marketing functions and more than 6,000 criteria, including writing a press release, creating a social post and writing email copy.
Optimizely’s models draw on Mark-IQ, the Agent Platform’s data layer, which is built on each organization’s own context, including experimentation history and web analytics, so every model has the brand’s context without a marketer having to rebuild it with every prompt. Using Mark-Bench’s all-pass rate on default configurations, Optimizely said its Agent Platform scored 67%, compared with 60% for Claude Code, at half the cost.
“The problem with generic harnesses is that they’re good for general use and productivity but very inefficient when it comes to domain-specific work,” said Shafqat Islam, president of Optimizely. “Marketing is a unique domain that requires agent harnesses to be specialized. By delivering frontier quality at lower costs, our customers finally have a way to scale agentic marketing to meet real-world demands rather than getting stuck in pilot purgatory.”
The announcement follows Optimizely’s launch of Virtual Teammates on Monday, giving marketers both the specialized models and role-specific AI co-workers to put them to work.
Also Read: Delight.ai’s John Kim on Building an AI That Remembers
As usage scales, selecting the right model for each task rather than defaulting to a single general-purpose model keeps costs predictable across large volumes of agentic work, the company said. Optimizely’s models are coming soon to all Agent Platform customers. More information is available at optimizely.com/products/ai.
