Warp
Agentic terminal for developers
About Warp
Warp is a terminal that was rebuilt rather than themed. The traditional terminal is a fifty year old interface that most developers use daily and few would design that way today: no block structure, no real editing, output and input interleaved into one scrolling stream. Warp treats each command and its output as a block you can select, share and rerun, gives you an input field that behaves like a text editor, and puts an AI agent in the same window.
The agent is the part that has changed the product. Rather than a chatbot bolted onto a shell, Warp Agent runs commands, reads their output, and works through multi-step tasks in the place where the work already happens. That matters because the terminal is where the context lives: the error you just hit, the branch you are on, the container that will not start. An assistant in a browser tab has to be told all of that, and an assistant in the terminal can see it.
More recently it has grown beyond one machine. Cloud agents run tasks off your laptop, and integrations with GitHub, Slack and Linear connect the terminal to where the rest of the work is tracked. The free tier is genuinely usable, including the option to bring your own AI inference rather than buying credits, which is unusual and a real cost saving if you already pay a model provider.
You install Warp and use it as your terminal. Everything a normal shell does still works, so adoption costs nothing beyond changing the app you open.
Commands and their output become blocks. You can scroll to a block from an hour ago, copy just its output, rerun it, or share a permalink to it with a colleague. The input line supports selections, multiple cursors and proper editing rather than readline shortcuts.
For AI work you prompt the agent in the same window. It can propose a command, run it, read the result and continue, so a task like tracking down why a build fails becomes a conversation with something that can actually see the failure. Usage is metered in credits, and you can point it at multiple LLM providers or supply your own key.
Cloud agents extend that to work that should not block your machine, and Warp Drive stores commands and workflows the team shares.
- •Block-based terminal - each command and its output is a selectable, shareable, rerunnable unit rather than an undifferentiated scroll of text.
- •Warp Agent in the terminal - an agent that runs commands and reads their output where the context already is, instead of being told about it in a browser tab.
- •Bring your own inference - use your own model provider keys on the free tier rather than buying credits, which most competitors do not allow.
- •Cloud agents - run longer tasks off your machine, with extended access on paid tiers.
- •Team workflows - Warp Drive shares commands and notebooks across a team, with unlimited collaboration on paid plans.
- •Integrations - GitHub, Slack and Linear, so terminal work connects to where issues and discussion live.
Developers who live in a terminal are the obvious audience, and the block model is worth the switch even before the AI features. If you regularly scroll back hunting for the output of a command you ran twenty minutes ago, that problem simply stops existing.
Infrastructure and platform engineers get the most from the agent, because their work is disproportionately made of commands whose output determines the next command. That loop is what the agent automates.
Teams benefit from the shared workflow layer, though the per-seat Business tier at $50 is a real cost that needs justifying against a free tier that is already good.
It is less compelling if your work is mostly in an IDE and the terminal is somewhere you occasionally run a test. The value scales with how much of your day is spent at a prompt.
Pricing
$0 - $200/mo
- Free$0/mo
- Build$20/mo
- Turbo$200/mo
- Business$50/mo
- EnterpriseContact sales
From the vendor pricing page, 2026-09-22
