Together Link: open models in the harness you already use. Start with one command today.

Together AI has announced Together Link, a tool that connects the coding agent harness a team already uses to open models served on Together AI. According to the company, Together Link supports Claude Code, Claude Desktop, Codex in the ChatGPT app and the CLI, OpenCode, and Pi. Together AI states that settings and logins stay as they were, and returning to native closed models takes one command.

Setup uses the command curl -fsSL https://link.together.ai/install | bash. After running it, the agent opens as before but is connected to Together AI using a Together AI API key; an account is required to obtain a key.

Together Link includes an "Auto" option that uses the Together AI Router, which reads each session's first task and selects a model. According to Together AI, quick fixes go to fast, low-cost models and hard problems get frontier capability. With an Anthropic key, Auto routes between Opus 5.5 and GLM 5.3; without one, it routes between GLM 5.3 and GLM 5.3 Flash. Routing occurs once per session, which Together AI says keeps prompt caching working.

A per-session tracker shows what was spent alongside what the same session would have cost on Opus 5.5. Billing runs on the existing Together API key against serverless pay-as-you-go or credit packs, with no separate contract required.

Together AI says Together Link runs on its serverless inference infrastructure, the same infrastructure developers choose for these models on OpenRouter. The company reports serving the largest share of OpenRouter tokens for DeepSeek V4.1 Flash (40.8%), GLM 5.3 Flash (28.2%), and Kimi K3 (23.1%), as of a 9/30/2026 footnote. Together AI also states that engineering organizations spend from tens of thousands to millions of dollars a month on closed models, and that open models such as Kimi K3, GLM 5.3, GLM 5.3 Flash and DeepSeek V4.1 Flash handle coding tasks at a fraction of the per-token price.