As enterprises grow more conscious of what AI actually costs, Writer is betting that efficiency — not raw benchmark scores — is the battleground.
On Thursday the company, which sells AI tools and agents for marketers, launched Palmyra X6, a new flagship model built as a post-training variation of Z.ai's open-source GLM-5.2. Combined with an upgraded "agentic harness" — the scaffolding that guides a model through multi-step tasks — Writer says customers could see costs fall by up to 50% for basic tasks.
"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 bet is grounded in research from Writer's own scientists. In a recent paper, the team tested small changes in harness efficiency across multiple models and found that harness tweaks were often a more reliable lever for cutting cost than the choice of model itself — with costs falling by an average of 40% across their testing.
"The harness is the one component whose efficiency multiplies across every model an organization runs — present and future," the researchers wrote.
For Writer's clients the experience remains model-agnostic: Palmyra X6 sits alongside other Writer models, or outside models imported through Azure or Amazon Bedrock. But Habib sees the cost push as part of a broader shift in the market — a growing distrust of major AI labs that, she argues, have a financial incentive to drive up token use.
"The cost explosion here is just unprecedented for customers, and so is the degree to which CIOs are giving up on the labs," she said, adding that the big labs "don't deeply understand right now how to help an enterprise get benefit from AI."
Both the model and the upgraded harness became available to Writer customers on Thursday.
