San Francisco enterprise AI company Writer today launched Palmyra X6, a new flagship model paired with a rebuilt agent harness that the company says cuts enterprise task cost by 52% and speeds task completion 48% versus the previous Palmyra generation. The release lands the same week Skan AI raised $63M for enterprise-agent grounding, underscoring the industry pivot toward orchestration economics.
Palmyra X6 Post-Trained On Z.ai's GLM 5.2
Writer built Palmyra X6 as a post-training variation of Z.ai's open-source GLM 5.2, priced at $2 per million input tokens and $8 per million output tokens (about $3.50 blended at a typical 3:1 mix). Per-task cost drops from roughly $0.25 to $0.12 versus the previous generation with no measured quality regression. On Writer's nine-benchmark evaluation, X6 scored 0.87 on average — ahead of Claude Opus 4.8 (0.86), GPT-5.5 (0.80), and Gemini 3.1 (0.77).
Eight-Hour Autonomous Runs
The model completes tasks in 26 seconds on average, produces 82 tokens per second, and can work unattended toward a long-horizon goal for up to eight hours — a coherence budget Writer designed for marketing, revenue, and operations workflows. The company also positioned Palmyra X6 as the least politically biased model it evaluated, an angle aimed at marketing teams needing tone-neutral content.
Harness As The Real Cost Lever
Writer's rebuilt agent harness — the orchestration layer that plans tasks, batches work, delegates to sub-agents and manages context — cuts costs 41% and completes tasks 44% faster across third-party models from Anthropic and OpenAI, not just Palmyra. CTO Waseem AlShikh: “The enterprise wants token consumption to explode — it means adoption is happening — but they need costs to flatten.” See our Anthropic inference-cost coverage.
Reporting based on coverage from SiliconANGLE, TechCrunch, and VentureBeat.