Enterprise AI bills are getting out of hand, and Writer just made its move to fix that. On Thursday, the company, which builds AI tools and agents for marketers, launched a new flagship model called Palmyra X6 aimed squarely at cutting deployment costs, according to TechCrunch AI. Paired with upgrades to Writer’s agentic harness, the company estimates customers could see costs drop by as much as 50% on basic tasks.
What stands out here isn’t just a new model. It’s the bet Writer is making about where the real savings live.
What Writer actually shipped
Two things went live to Writer clients on Thursday, as detailed in TechCrunch AI:
- Palmyra X6: a new flagship model built as a post-training variation on Z.ai’s open source GLM-5.2. Writer says it delivers deployment-ready capabilities at a much lower price point.
- Harness upgrades: significant improvements to Writer’s standard agentic harness, the infrastructure that decides how a model plans, calls tools, and works through a task.
The combined pitch targets complex, multi-step jobs, the kind that eat tokens fast. Writer says its new approach runs those tasks quicker and with fewer tokens, which is where the cost reduction comes from.
Why the harness matters more than the model
Here’s the part practitioners should pay attention to. Writer isn’t only selling a cheaper model. It’s arguing that the harness is the bigger lever.
A recent paper from Writer’s researchers tested small changes in harness efficiency across several different models. They found that tweaking the harness was often a more reliable way to cut costs than swapping the model itself, with expenses falling an average of 40% in their testing.
“The harness is the one component whose efficiency multiplies across every model an organization runs, present and future,” the researchers wrote, according to TechCrunch AI.
That’s a meaningful claim. Most cost conversations fixate on which model to pick and what it charges per token. Writer’s data suggests the plumbing around the model can matter just as much, and it applies no matter what model you’re running.
The context: everyone’s chasing cheaper tokens
The backdrop is a broader shift in how companies think about AI spend. Open source models offer far lower per-token costs, but finding the right one for a specific job is hard. Palmyra X6 is Writer’s attempt to package that value into something enterprises can deploy without the guesswork.
Writer also keeps the experience model-agnostic. Palmyra X6 sits alongside Writer’s other models or outside models pulled in through Azure or Amazon Bedrock, so clients aren’t locked into one engine.
CEO May Habib framed the launch as a direct answer to enterprise frustration. “I think the enterprise is absolutely sick of chasing the next benchmark,” she told TechCrunch AI. “They want flattening cost, and it seems like nobody can deliver that.”
Habib went further, tying the cost push to growing distrust of major AI labs, which 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 labs “don’t deeply understand how to help an enterprise get benefit from AI.”
Why it matters
This is significant because it reframes the cost debate. If Writer’s research holds up, teams obsessing over per-token pricing might be optimizing the wrong variable. The harness, the layer most buyers never think about, could be the cheaper win.
For anyone running AI at scale, the takeaway is practical: audit how your agents execute tasks, not just which model powers them. Expect more vendors to lean into harness efficiency and open source post-training as the cost pressure keeps building.
Writer’s model and harness upgrades are available now. Full details are at the original TechCrunch AI report.