# Action-Selecting AI Models Stanford and Nvidia Has Built the CLM-8B Contrastive Language Model By Adam Harrie | Written with AI assistance | Published 2026-09-25 | Updated 2026-09-28 | Tech Source: Trend Hunter, https://www.trendhunter.com/trends/action-selecting-ai-models References: [venturebeat](https://venturebeat.com/technology/stanford-and-nvidias-open-clm-8b-caches-reusable-agent-actions-and-runs-up-to-9x-faster-than-jev-in-tests) & [contrastive-lm.notion.site](https://contrastive-lm.notion.site/)  Stanford and Nvidia built 'CLM-8B' so [AI agents](https://www.trendhunter.com/trends/Agent-ready-development-platforms) can choose from a bounded set of actions without generating a full-text reply. The contrastive language model (CLM) encodes the current state and candidate actions, then selects the closest match for faster routing and ranking within workflows. Built on a frozen Qwen3-8B backbone with separate heads for states and actions, the model can cache recurring action embeddings rather than recomputing them each time. In zero-shot tests, it ran up to 9x faster than TypeSafe's Jev, especially when the same options were reused across many requests. For providers, CLM-8B offers a dedicated decision layer for tasks like [tool routing](https://www.trendhunter.com/trends/toolhouse1) or ticket triage, where the menu of choices remains stable. That could reduce latency across long agent chains and leave [larger reasoning models](https://www.trendhunter.com/trends/Specialized-AI-models) focused on generating answers rather than repeatedly choosing among them. Image Credit: CLM Team ## Trend Insights (Trend Hunter) - Score: 6.6/10 - Popularity: 54% | Activity: 46% | Freshness: 99% - Audience gender: 50% men, 50% women - Primary generations: Millennial, Gen X - Top markets: North America ## Categories [Trend Hunter](https://www.trendhunter.com/trends) > [Tech](https://www.trendhunter.com/tech) > [AI](https://www.trendhunter.com/ai) ## Key Themes ### What's Driving This Trend - **Bounded Action Models:** AI systems that choose from predefined options create room for lower-latency agent workflows where routine decisions no longer require full generative responses. - **Cached Action Embeddings:** Reusable action representations introduce efficiency gains for high-volume enterprise tasks that rely on stable menus, queues, or tool sets. - **Agent Decision Layers:** Dedicated routing models separate selection from reasoning, enabling modular AI stacks that optimize cost, speed, and accuracy across complex workflows. ### Who This Affects Most - **Enterprise Software:** Workflow platforms can incorporate faster AI routing to streamline ticket triage, task assignment, and tool selection inside existing business systems. - **Customer Support:** Support operations benefit from bounded-choice intelligence that classifies issues and routes cases without slowing down escalation or response pipelines. - **Cloud AI Infrastructure:** Model providers have new opportunities to offer specialized decision services that reduce compute demand across repeated agentic actions. ## Related on Trend Hunter - [Vertical LLMs](https://www.trendhunter.com/trends/base-model.md) - [Adaptive Reasoning Models](https://www.trendhunter.com/trends/adaptive-llm-efficiency.md) - [Agentic AI Models](https://www.trendhunter.com/trends/nvidia-nemotron-models.md) - [Advanced Language Model Series](https://www.trendhunter.com/trends/qwen3.md) - [Multi-Model Answer Platforms](https://www.trendhunter.com/trends/collectiviq-by-buyers.md) - [Long-Horizon AI](https://www.trendhunter.com/trends/Long-horizon-AI.md) - [Large Language Models](https://www.trendhunter.com/trends/large-language-models.md) - [Advanced AI Models](https://www.trendhunter.com/trends/deephermes-3.md) - [AI Routing Tools](https://www.trendhunter.com/trends/focal-ai.md)