Writer introduces new AI model and upgraded harness to contain token costs

The AI industry is witnessing growing concerns over the escalating costs associated with model deployments, prompting users to seek more affordable solutions. Writer, a provider of AI tools and agents for marketers, has introduced a new flagship model named Palmyra X6, designed to address this challenge. This model is based on a post-training variation of Z.ai's open-source model GLM-5.2 and promises to deliver deployment-ready capabilities at a significantly lower price point.

Writer claims that the combination of Palmyra X6 and upgraded harness infrastructure could slash costs for its customers by up to 50% for basic tasks.

In addition to the new model, Writer has rolled out substantial enhancements to its standard agentic harness. These improvements are aimed at optimizing performance, reducing token costs, and enabling faster execution of complex, multi-step tasks. The new harness optimization strategy emphasizes efficiency gains that can be consistently applied across various models, rather than relying solely on model selection.

A research paper by Writer's team supports this approach, demonstrating that harness improvements can lead to an average 40% reduction in costs compared to model changes.

The launch of Palmyra X6 positions Writer to offer a model-agnostic solution to its clients, with the ability to integrate with other models or those imported through Azure or Amazon Bedrock. However, CEO May Habib emphasizes that the cost-cutting push is also fostering a growing skepticism towards major AI labs, which have a vested interest in increasing token usage.

Habib notes that enterprises are increasingly questioning the effectiveness of these labs in delivering tangible benefits from AI, as they fail to grasp the intricacies of optimizing costs for their customers.