.png){kind=logo}
Writer has launched Palmyra X6, a new flagship AI model built for enterprise marketing and revenue tasks. The company also upgraded its agent platform to reduce token use during complex workflows.
The model and platform updates became available to Writer customers on Thursday. Writer says the combined system lowers average costs while improving speed and task quality.
Writer developed Palmyra X6 through post-training work on Z.ai’s open-source GLM-5.2 model. GLM-5.2 uses a mixture-of-experts structure that activates selected parts of the model for each request.
The company designed X6 for content creation, brand voice controls, sub-agent delegation and long workflows. Writer charges $2 per million input tokens and $8 per million output tokens. It says basic enterprise tasks could cost up to 50% less under the new setup.
Chief Executive May Habib said companies now care more about practical costs than benchmark results. “I think the enterprise is absolutely sick of chasing the next benchmark,” she said. “They want flattening cost, and it seems like nobody can deliver that.”
Meanwhile, Writer reports that X6 scored an average of 0.87 across nine internal evaluations. The company compared this result with Claude Opus 4.8, GPT-5.5 and Gemini 3.1. Those models scored 0.86, 0.80 and 0.77, respectively.
Writer also updated the harness that manages how AI agents receive instructions, use tools and complete tasks. The company says the upgraded system reduces repeated steps and limits unnecessary token use.
When paired with Palmyra X6, the Writer Agent platform operates at an average cost reduction of 52%, according to company figures. Writer also reports a 48% speed increase and a 10% quality improvement.
A Writer research paper examined harness efficiency across several AI models. The researchers found that small harness changes reduced average costs by 40% during testing. In many cases, these changes produced more reliable savings than switching models.
“The harness is the one component whose efficiency multiplies across every model an organization runs—present and future,” the researchers wrote. Writer customers can still use other company models or import outside models through Azure and Amazon Bedrock.
Writer says Palmyra X6 completes tasks in 26 seconds on average and generates 82 tokens per second. The model can also work toward a long-term goal for up to eight hours without human input.
The company added reporting and governance controls to its agent platform. Administrators can now track activity, spending and performance across teams. They can also compare usage over selected periods.
Employees can create and share Playbooks and Skills through Writer Agent. Playbooks provide reusable workflows, while Skills contain repeatable instructions for common tasks. Administrators can manage ownership, status and usage from one dashboard.
Moreover, companies can set consumption limits and spending alerts for individual tasks or Skills. These controls give leadership teams clearer data on adoption and operating costs.
Writer co-founder and Chief Technology Officer Waseem AlShikh linked rising token use with wider adoption. “The enterprise wants token consumption to explode—it means adoption is happening—but they need costs to flatten,” he said.
Also Read: Most Powerful DeepSeek Prompts Every Writer Should Use
