Writer shipped Palmyra X6, its new flagship model, alongside an upgraded WRITER Agent and rebuilt governance tools in AI Studio. Together they target enterprises that need agentic workflows in production without AI spending that outpaces the business case.
The package addresses a specific operational failure: enterprises running through annual AI budgets within months, with uncapped model spend reaching upward of $500 million monthly at scale. Writer prices Palmyra X6 at $2.00 per million input tokens and $8.00 per million output tokens, roughly $3.50 blended at a 3:1 mix. Per-task cost falls from roughly $0.25 to $0.12 versus the prior generation, with latency dropping in step. The model can sustain coherent, unattended reasoning for up to eight hours on a single goal, the window the company says makes long background automation financially practical.
The updated WRITER Agent runs on Palmyra X6 or any other model in the platform, including third-party ones. Across all tested models, Writer reports the new version completes tasks 44% faster and at 41% lower cost per task. Paired specifically with Palmyra X6, those figures improve further: 52% lower cost, 48% faster completion, and a 10% quality gain. The agent now adjusts its reasoning approach based on task complexity: direct answers for simple requests, structured plans for multi-step work. It can also distribute high-volume tasks across sub-agents rather than processing them sequentially, surfaces outputs from connected tools including Outlook drafts and Snowflake data tables directly in the session interface, and prompts users with structured clarifying questions before executing ambiguous instructions.
The AI Studio overhaul, which Writer redesigned for teams managing agent adoption at scale, closes the administrative gap that widens when pilots expand company-wide. Admins get a centralized Home view covering sessions, playbook runs, active users, token usage, and spend trends, filterable by team or billing group. Hard consumption limits and real-time alerts let teams cap overuse before it affects the invoice. Multi-model support now extends to WRITER Agent: administrators can enable third-party models from Azure, AWS Bedrock, and NVIDIA NIM for managed teams, assigning specialized models to specific task types including image generation.
Palmyra X6, the upgraded WRITER Agent, and the new AI Studio governance features are all available starting today. The performance figures cited are Writer's own benchmark results and have not been independently verified.
