Writer releases Palmyra X6 and upgrades its agentic harness

Writer has launched Palmyra X6, a new flagship AI model, alongside major upgrades to its standard agentic harness. The company estimates that the model and infrastructure changes together can reduce customer costs for basic tasks by as much as 50%. Both capabilities became available to Writer clients on Thursday.

Palmyra X6 is a post-training variation of Z.ai’s open-source GLM-5.2 model. Writer, which provides AI tools and agents for marketers, positions the release as a deployment-ready option intended to lower the cost of enterprise AI use.

Model and harness are released together

The company’s approach focuses on complex, multi-step tasks. Writer says the new system is intended to execute such work faster and with fewer tokens, making the surrounding harness infrastructure a central part of the cost equation rather than treating model selection as the only lever.

For customers, the platform remains model-agnostic. Palmyra X6 will sit alongside Writer’s other models and external models imported through Azure or Amazon Bedrock. This arrangement allows clients to use the new model without making it the sole model available in their deployments.

Research points to workflow efficiency

A paper from Writer researchers tested small changes in harness efficiency across several models. The research found that harness changes were, in many cases, a more reliable way to reduce costs than choosing a different model, with costs falling by an average of 40% across the tests.

“The harness is the one component whose efficiency multiplies across every model an organization runs—present and future,” Writer’s researchers wrote.

Writer CEO May Habib said enterprises are focused on reducing costs rather than continually pursuing the next benchmark. Her comments reflect a wider concern about the cost of token-intensive AI deployments and the ability of major AI labs to help enterprises obtain value from those systems.

Implications for enterprise AI choices

The release also arrives as the market weighs open and proprietary approaches, a choice explored in open AI and proprietary model market scenarios open AI and proprietary model market scenarios as organizations balance cost, capability and deployment requirements.

For businesses, Writer’s figures make a practical case for assessing the full operating path of an AI task: token use, multi-step execution and harness design, as well as the underlying model’s price. Teams selecting models through Writer, Azure or Amazon Bedrock can use that assessment to identify where efficiency work may produce measurable cost reductions.