Meet Together Link: A Free CLI That Runs Open Models Like Kimi K3 and GLM 5.3 Inside Claude Code, Codex, an…

What it does

Together AI launched Together Link, a new free command-line interface (CLI) tool designed to connect popular AI interfaces like Claude Code, Codex, OpenCode, Pi, and desktop apps for Claude and ChatGPT to open-source models such as Kimi K3 and GLM 5.3. A single install command sets up the CLI. It includes an Auto router that dynamically selects the best open model for each session, streamlining access across different platforms without manual switches. Together Link is released under the permissive MIT license.

Why it matters

Together Link reduces friction for operators and developers aiming to deploy open AI models alongside proprietary clients. Many users rely on Claude, Codex, or ChatGPT desktop apps but want to test or run open models without juggling multiple tools. This CLI simplifies that by bridging open and closed environments seamlessly. The Auto router feature smartly optimizes which model runs each session, improving efficiency. Together claims this approach cuts operational costs by over 50 percent, signaling potential savings that matter for startups, small teams, or cost-conscious devs experimenting with large language models.

Who it is for

The tool primarily targets AI builders and operators who need quick, flexible access to both proprietary AI apps and open AI models on the command line. Founders or developers working on research, prototypes, or applications that mix Claude and OpenCode environments will find this especially useful. It also benefits businesses keen on exploring cheaper open models without abandoning popular AI desktop clients or workflows. This lowers barriers to testing or adopting open models for those accustomed to polished apps.

The catch

Though Together Link offers a convenient interface and cost savings, open models like Kimi K3 and GLM 5.3 typically perform differently from proprietary models, sometimes with lower accuracy or speed. The claim of 50 percent savings depends on the actual use case and infrastructure. Users should cautiously benchmark results and consider if integration complexity or model trade-offs affect application goals. Also, as a new free tool, community and support levels might be limited initially.

What to watch next

Watch for wider adoption and real-world benchmarks from builders using Together Link in production scenarios. The evolution of the Auto router’s model selection logic will be key to performance and cost optimization. Together’s plans to support additional open models and integrate with more AI apps will determine whether this CLI becomes a go-to flexible AI orchestration tool. Also track whether this spurs competitors or inspires similar open-source bridging solutions for hybrid AI workflows.

AI Quick Briefs Editorial Desk