Writer launches Palmyra X6 and a harness upgrade to cut AI token costs

Across the AI industry, users are growing more conscious of how expensive their deployments can be, and feeling new urgency to cut costs — but while open source models offer lower per-token costs, finding the right one for a given job can be difficult. On Thursday, Writer, which builds AI tools and agents for marketers, launched a new flagship model called Palmyra X6 to address that problem. Built as a post-training variation on Z.ai’s open source model GLM-5.2, Writer says the system delivers deployment-ready capabilities at a much lower price, and estimates the model, combined with changes to its harness infrastructure, will cut costs for customers by as much as 50% on basic tasks. Both the model and harness upgrades became available to Writer clients starting Thursday.

“I think the enterprise is absolutely sick of chasing the next benchmark,” CEO May Habib told TechCrunch. “They want flattening cost, and it seems like nobody can deliver that.” The new approach emphasizes complex, multi-step tasks executed faster and with fewer tokens, treating harness optimization as a crucial lever for getting there. A recent paper from Writer researchers found that, across multiple models tested, changes to the harness were often a more reliable way to cut costs than switching models, reducing costs by an average of 40% in their testing — evidence, the researchers wrote, that the harness is the one component whose efficiency gains multiply across every model an organization runs, present and future.

For Writer’s clients, the experience remains model-agnostic: Palmyra X6 sits alongside Writer’s other models and outside models imported through Azure or Amazon Bedrock. Habib also sees the cost-cutting push as part of a broader shift away from trusting major AI labs, which she said have a financial incentive to drive up token use and, in her view, don’t fully understand how to help an enterprise get value from AI.

Why it matters

If harness efficiency, not raw model capability, is what actually drives enterprise AI costs down, that shifts competitive pressure away from the frontier labs racing to build the biggest models and toward the vendors optimizing how those models are actually used.