GKRootWire
Gadgets HP's Convertible OmniBook X Flip Drops to $699 at Best BuyGadgets Xiaomi 17 Ultra's 'Moon Mode' Got Fooled by a Solar EclipseAI Cerebras Partners with OpenAI to Supercharge GPT-5.6 Inference SpeedAI Lumabri Lets You Run Mixture-of-Experts AI Models Across a P2P SwarmDev Tools ArcadeMaker Brings a Custom Scripting Language and IDE to C# Game DevelopmentDev Tools DeepSeek Launches 'Harness' Developer Preview for Agentic CodingGadgets HP's Convertible OmniBook X Flip Drops to $699 at Best BuyGadgets Xiaomi 17 Ultra's 'Moon Mode' Got Fooled by a Solar EclipseAI Cerebras Partners with OpenAI to Supercharge GPT-5.6 Inference SpeedAI Lumabri Lets You Run Mixture-of-Experts AI Models Across a P2P SwarmDev Tools ArcadeMaker Brings a Custom Scripting Language and IDE to C# Game DevelopmentDev Tools DeepSeek Launches 'Harness' Developer Preview for Agentic Coding
AI

Writer Launches New Model and Harness to Slash AI Token Costs

The enterprise AI startup built its latest system atop Z.ai's open source GLM-5.2 model to cut deployment costs without sacrificing capability.

Writer, the enterprise-focused AI company, has released a new model paired with an upgraded orchestration harness designed to make deployment cheaper and more efficient. Rather than training a foundation model from scratch, Writer took Z.ai's open source GLM-5.2 model and applied its own post-training techniques to fine-tune it for enterprise use cases.

The accompanying harness—the software layer that manages how the model handles tasks, tool calls, and context—has also been upgraded specifically to reduce token consumption, which is often the biggest hidden cost of running AI agents in production. Writer is positioning the combined package as "deployment-ready," meaning businesses can plug it in without extensive additional tuning.

By building on an open source base rather than a proprietary one, Writer avoids licensing fees tied to closed models while still offering competitive performance, according to the company.

Why it matters: Token costs have become a major bottleneck for companies scaling AI agents beyond pilot projects, and this move signals a broader industry shift toward optimizing existing open models rather than always training new ones from scratch. It's also a sign that open source foundation models like GLM are becoming credible bases for commercial products, not just research curiosities.

Sources: TechCrunch