Crush
Glamourous agentic coding in your terminal, wired to your own LLM of choice.
Crush is a serious, free alternative for terminal-centric developers who want multi-model AI coding without monthly fees. Its LSP and MCP integration set it apart, but it requires manual setup and API keys. If you live in the CLI and value control, Crush is worth adopting; if you prefer a GUI or zero-config, look elsewhere (e.g., GitHub Copilot or Cursor).
Verified 7d ago · liveness 87/100 · cite: rightaichoice.com/tools/crush
- Terminal-loving developers who want an AI assistant integrated into their CLI workflow
- Users who need multi-model flexibility to compare or switch LLMs per task
- Developers working on headless servers or via SSH who need AI assistance in the terminal
- Power users who want extensibility via MCPs and custom LSP integrations
- Users seeking a graphical, IDE-like AI coding assistant with no terminal use
- Non-developers or beginners who want a zero-config, cloud-hosted AI coding tool
- Teams requiring centralized billing, usage controls, or admin dashboards
We scan live Reddit threads, YouTube comments, X posts, G2 reviews and other communities — and hand you an honest verdict in under a minute.
- Honest verdict, not marketing
- Real pros & cons from real users
- Attributed quotes with receipts
3 free scans · no card needed
Skip Crush if you need a zero-config, graphical AI coding assistant, are a beginner uncomfortable with the terminal, or require centralized billing and admin controls for a team.
You pay for API usage directly with your chosen LLM provider, so costs scale with your usage and can exceed a flat subscription if you frequently use premium models like GPT-4o.
Crush is free and open-source, so you only pay for your own API usage with providers like OpenAI or Anthropic. This beats subscription-based assistants like GitHub Copilot ($10/mo) or Cursor Pro ($20/mo) if you're a low-volume user, but heavy usage could rival those costs. It's ideal for individuals and small teams that want cost control and flexibility.
In short
Crush — Glamourous agentic coding in your terminal, wired to your own LLM of choice. Best for Terminal-loving developers who want an AI assistant integrated into their CLI workflow, Users who need multi-model flexibility to compare or switch LLMs per task, Developers working on headless servers or via SSH who need AI assistance in the terminal. Free to use.
Viability Score
How well maintained and how widely used is Crush? Built from what the vendor actually publishes (docs, changelog, tutorials, integrations, pricing), whether the site is live, and how much real users discuss it. How we calculate this
Last calculated: September 2026
How we score →Key Features
- Multi-model support (OpenAI, Anthropic, Groq, OpenRouter, Vercel AI Gateway)
- Switch LLMs mid-session while preserving context
- Session-based workflows for multiple contexts per project
- LSP-enhanced code understanding (gopls, nil, etc.)
- Extensible via MCPs (http, stdio, sse)
- Cross-platform: macOS, Linux, Windows, Android, FreeBSD, OpenBSD, NetBSD
- Built on Charm ecosystem (25k+ apps)
- Install via Homebrew, npm, Arch, Nix, Winget, Scoop, Debian, Fedora, Go
- Open-source (MIT license) on GitHub
- Environment variable configuration for API keys
- NixOS and Home Manager module support via NUR
- Compact TUI mode and debug options
- Custom providers via OpenAI- or Anthropic-compatible APIs
- Experimental: .agents folder for custom agents
- Command approval gates for shell execution safety
About Crush
Crush is a free, open-source, terminal-native AI coding assistant from Charmbracelet. It's built for developers who live in the CLI, letting you bring agentic coding to your favorite terminal on macOS, Linux, Windows (PowerShell and WSL), Android, FreeBSD, OpenBSD, and NetBSD. You wire your own tools, code, and workflows into the LLM of your choice, paying only for API usage with your chosen provider—a cost-effective alternative to subscription-based cloud assistants. At its core, Crush is flexible. It supports a wide range of major LLMs—including OpenAI, Anthropic, Groq, OpenRouter, and Vercel AI Gateway—and lets you add custom providers via OpenAI- or Anthropic-compatible APIs. You can switch between models mid-session without losing context, which is ideal for comparing outputs or tailoring model choice to the task. Session-based workflows let you maintain multiple contexts per project, so you can keep different threads of work separate and organized. Crush enhances code understanding through Language Server Protocol (LSP) integration, using language servers like gopls for Go or nil for Nix to provide additional context, much like a human developer would. This gives the assistant deeper awareness of your project's structure and symbols. Extensibility is a key strength: Crush supports MCPs via http, stdio, and sse, letting you plug in external tools and data sources. It also includes command approval gates for shell execution safety, and an experimental .agents folder for defining custom agents. For Nix users, Crush offers first-class support with NixOS and Home Manager modules through NUR, declarative configuration options, and simple installation via flake. The tool is built on the Charm ecosystem, which powers 25k+ applications, ensuring industrial-grade reliability. Compared to cloud-based assistants like GitHub Copilot or Aider, Crush gives you full control over your LLM choices and data flow, all within a terminal.
Behind the Verdict
Crush shines among terminal-based AI coding tools by putting you in control. Unlike Aider or GitHub Copilot, which lock you into their default model or workflow, Crush lets you switch between OpenAI, Anthropic, Groq, OpenRouter, and more mid-session without losing context. This is a genuine productivity win when you need to compare model outputs or match a model to a specific task. The LSP integration is a differentiator: by hooking into language servers like gopls (Go) and nil (Nix), Crush gains project-aware context that many CLI assistants lack. Combined with MCP support, you can extend it with your own tools and data sources—something you can't do with closed assistants. Where Crush struggles is in the GUI-driven workflows. If you prefer an IDE-like experience with diff reviewing in a visual interface, Crush's TUI is limiting. And while it's open-source and free, you'll need to manage your own API keys and billing across providers, which can be a hassle for teams. There's no centralized billing or admin dashboard here. For NixOS users, the declarative module support is a huge plus, letting you configure Crush as part of your system config. It's also lightweight and works over SSH, making it a go-to for headless dev servers. If you're a terminal purist who wants multi-model flexibility and doesn't mind manual setup, Crush is a solid pick. If you need a zero-config, GUI-driven assistant for a team, consider GitHub Copilot or Cursor instead.
Researching Crush? Get your full AI stack in 60 seconds.
Free, no signup — tell us your goal and get tools matched to your budget & existing stack.
Real-world workflow fit
Concrete scenarios for the personas Crush actually fits — and what changes day-one when you adopt it.
You're working on a Go project and want AI help without leaving the terminal. You install Crush via Homebrew, configure your OpenAI API key, and use gopls LSP for context.
Outcome: You get inline code completions and diagnostics right in your terminal, using Crush to refactor functions and generate tests, all without switching windows.
You manage your system with Nix and want Crush integrated into your configuration.
Outcome: You add the NUR module, set up your providers and LSP servers declaratively, and Crush is available across your system with your preferred settings.
You SSH into a headless server to work on a web app, and you need AI assistance.
Outcome: You run Crush via tmux on the server, connect it to Groq for fast responses, and use MCP to access the project's database schema, making changes efficiently.
Use Cases
- Run an AI coding assistant over SSH on a remote dev server without an IDE.
- Work inside tmux panes alongside your editor and shell, all in the terminal.
- Swap between GPT-4o, Claude Sonnet, and a local Llama model in the same session.
- Use LSP integration to get code diagnostics and inline completions in the terminal.
- Execute shell commands via the assistant with approval gates for safety.
- Define custom agents for specialized tasks using .agents folder.
- Maintain multiple session contexts for different features or bug fixes.
- Extend Crush with MCP servers for external tools and data.
Models Under the Hood
as of 2026-08-31
Limitations
- TUI has inherent UI limits — reviewing a 500-line diff or a visual design spec is awkward compared to an IDE.
- Younger project than Aider or Claude Code — expect rough edges on the long tail of commands.
- Shell execution requires careful approval-gate setup for safety.
- No centralized team management or usage dashboards.
as of 2026-08-30
Verification history
We have re-verified Crush 18 times since . Each pass re-reads the vendor's own pages and re-checks every listed field against that evidence; passes where nothing had changed are marked as such.
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
Showing the 6 most recent of 18 verification passes.
Free to cite with attribution — this page re-verifies continuously.
12-month cost
Project the real annual outlay, including the implied monthly cost when only an annual tier is published.
Vendor list price only. Add-on usage, seat overages, and contract minimums are surfaced under Hidden costs & gotchas.
Plans compared
For each published Crush tier: who it actually fits, and what it adds vs. the previous tier. Cross-reference the cost calculator above for projected annual outlay.
Open Source
$0/mo
Ideal for
Individual developers and hobbyists who want a free, self-hosted terminal AI assistant and are comfortable managing their own API keys and providers.
What this tier adds
The starting tier, free forever, includes the full CLI tool with multi-model, LSP, and MCP support, with no subscription fees.
Hyper
Subscription with free tier
Ideal for
Users who want a managed, privacy-focused provider with zero data retention and GDPR compliance, without setting up their own API keys.
What this tier adds
Adds the official Crush provider as a managed option, with a free tier available and privacy guarantees—unlike the Open Source tier where you bring your own provider.
Where the pricing makes sense
The company stage and team size where Crush's pricing actually pencils out — and where peers do it cheaper.
Crush is free and open-source, so you only pay for your own API usage with providers like OpenAI or Anthropic. This beats subscription-based assistants like GitHub Copilot ($10/mo) or Cursor Pro ($20/mo) if you're a low-volume user, but heavy usage could rival those costs. It's ideal for individuals and small teams that want cost control and flexibility.
Setup time & first value
How long it actually takes to get something useful out of Crush — broken out by persona, not the marketing-page minute.
For a Homebrew or npm install, you can be up and running in under 10 minutes: install, set your API key via environment variable, and start your first session. Nix users may need 15-20 minutes to set up the NUR module, but a quick nix-shell gets you going in minutes.
Switching to or from Crush
How to bring data in from common predecessors and how to get it back out — written for the switcher, not the buyer.
- →From GitHub Copilot in terminal: Install Crush, point it to your preferred LLM, and start using its session-based workflows for a more flexible, multi-model experience.
- →From Aider: Configure your providers and LSP servers in Crush's config, and migrate your existing prompts or workflows to Crush's session model.
- ↗To GitHub Copilot: If you need a GUI or centralized team management, you can continue using Crush for terminal work and Copilot for IDE integration.
- ↗To Cursor: For a graphical, more autonomous experience, you can switch, exporting your session contexts manually.
Integrations
Resources & Guides
- Resourcegithub.com
GitHub - charmbracelet/crush: Glamourous agentic coding for all 💘
Glamourous agentic coding for all 💘. Contribute to charmbracelet/crush development by creating an account on GitHub.
- Resourcegithub.com
crush/README.md at main · charmbracelet/crush
Glamourous agentic coding for all 💘. Contribute to charmbracelet/crush development by creating an account on GitHub.
Tutorials & Learning
Official links
Tools that pair well with Crush
Common stack mates teams adopt alongside Crush, with the specific reason each pairing earns its keep.
Alternatives to Crush
View all
Poolside AI
Open-weight agentic coding models for secure on-prem enterprise AI
Amazon CodeWhisperer
AWS-native AI coding assistant with agentic automation for developers
GitHub Copilot
AI code completion and agentic coding with multi-model flexibility
Frequently Asked Questions
Best-of guides
Used Crush? Help shape our editorial sentiment research.