Claude now talks to any AI model — the hybrid team revolution

The news drops on GitHub, where a developer released cc-fleet — a tool that turns Claude into a sort of AI model project manager. The trick? Take any LLM — DeepSeek, GLM, Qwen, Kimi, MiniMax, whatever — and Claude integrates it as a real team member, not a secondary API.

How does it technically work? Claude Code (its superpower for writing code) accesses external models in real-time, using them both as stable teammates in the squad or as one-off subagents that handle specific tasks on demand. It's like having a boss who calls the right colleague for the right job, no hassle.

Why does it matter? First: it breaks the monopoly of single models. If DeepSeek is faster at reasoning but Qwen is better at natural language, now you can use both in the same workflow without integration headaches. Second: it opens the door to Chinese models — technically solid but historically sidelined — letting them actually compete in Western scenarios.

The economic angle is obvious: you're no longer forced to pay one API for everything. You use the cheapest or best model for each situation, cutting costs and (theoretically) boosting results. A real melting pot where marketing doesn't win — the right architecture does.

This tool represents a bigger shift: the monolithic ecosystem of single-model dominance is cracking. Claude isn't going anywhere, but its strength becomes orchestration, not being the only player.