xAI dropped Grok Build CLI. Google has agents-cli. Claude Code is a CLI. Aider is CLI. Theres a clear trend here.
CLI is becoming the default interface for AI coding agents. Makes sense. Terminal is where developers live. Lower latency than web UIs. Easier to pipe output. No electron overhead.
But each CLI optimizes for different stuff. Grok Build wants to be a general coding assistant tied to grok. Agents-cli is for deploying AI agents on gcloud with skill packs. Aider is pure git based editing. Claude Code is conversational.
Then you have verdent which skips CLI entirely and goes desktop app + vscode extension. Does the multi agent orchestration thing instead of single model chat. Different approach.
For local model people, the question is: which of these actually work with local models? Aider yes. Most others are locked to their own model. If youre running qwen or deepseek locally, your options are pretty limited right now.
I think well see more hybrid approaches eventually. CLIs that can route to local models for simple tasks and cloud models for complex ones. The skill pack idea from google is interesting because it separates the knowledge from the model. You could theoretically inject those skills into any model, local or cloud.
Right now I use aider with local models for quick stuff and cloud based tools for complex tasks. Works ok but the context switching is annoying. Wish there was one tool that handled both.
[link] [comments]