Writer launches major agentic AI improvements with Palmyra X6 flagship model

Writer has pushed out Palmyra X6, its new flagship model, and paired it with a heavier set of agent upgrades aimed squarely at marketing and revenue teams. The pitch is simple: get frontier-style performance without letting the bill run away.

The company says its Agent platform now runs at 52% lower cost on average when used with X6, while speed improves by 48% and quality by 10%. That matters because Writer is targeting workloads that are messy, repetitive and hard to automate cleanly — the kind of work where agents need to coordinate, hand off sub-tasks and stay on brand.

X6 is the follow-up to X5, which arrived in April 2025. Writer says it built the new model for enterprise content generation, sub-agent delegation and brand voice, then post-trained it on top of GLM 5.2, the open-source mixture-of-experts model from Z.ai Co. Ltd. On Writer’s own evaluation, X6 averaged 0.87 across nine tests at $2 per million input tokens and $8 per million output tokens, edging Claude Opus 4.8, GPT-5.5 and Gemini 3.1 on that scoring setup.

The company is also leaning hard into the long-haul side of agents. It says X6 completes tasks in 26 seconds on average, generates 82 tokens per second and can stay on a goal for up to eight hours without human intervention. Writer also says the model is its least politically biased one, a detail that fits its pitch to teams trying to keep output neutral enough to adapt to brand rules.

Behind the model, Writer is upgrading reporting and governance. Administrators get a clearer view of activity across teams, employees can build and share Playbooks and Skills, and the system now includes dedicated analytics for those workflows. Leadership can also set consumption limits and alerts at specific thresholds, which is usually the point where the sales deck stops sounding poetic and starts sounding like finance got a vote.