D

Command-line Assistants

Devin CLI

Cognition's local coding agent for the terminal, written in Rust. It works against your files, tools, and shell with any frontier model, and hands the same session to Devin in the cloud when a task outgrows your laptop.


Highlights

  • Hand off to the cloud

    /handoff moves the session to a Devin cloud agent with its own VM, browser, and tests, and you come back to a finished pull request.

  • Multi-model

    Pick from frontier models including Claude Opus 4.8, Claude Fable 5, GPT-5.5, and Cognition's own SWE-1.6 and SWE-1.7.

  • Fast terminal UI

    A custom Rust rendering library keeps the interface snappy; Cognition demoed it running on a 1978 VT100.

  • MCP and permissions

    First-party MCP support with the same allow, deny, and ask rules as the built-in tools.

Available Models

Claude Fable 5 Claude Opus 4.8 GPT-5.5 SWE-1.7 SWE-1.6

Platforms

macOS Linux Windows

Install with curl -fsSL https://cli.devin.ai/install.sh | bash, brew install --cask devin-cli, or irm https://static.devin.ai/cli/setup.ps1 | iex in PowerShell, then run devin in a project directory.

Pricing

Freemium
Free $0
Pro $20/month
Max $200/month
Team $80/month + $40 per seat
Enterprise Custom
CLI usage draws from the same daily and weekly allowance as the rest of your Devin plan. Free has a light quota and limited models; Pro adds full model access and Devin Cloud; extra usage on paid plans is billed at API pricing.