# OpenCode OpenCode is an open-source AI coding agent for terminal, IDE, and desktop, helping developers understand codebases, plan changes, and use multiple LLM providers. - Findings24: - Website: - Pricing: Freemium - Categories: AI Code Assistant, AI Developer Tools ## What is OpenCode? OpenCode is an open source AI coding agent for people who want to work with an agent in the terminal, an IDE, or a desktop app. The product is positioned as a general coding assistant that can help explain code, plan changes, and make edits inside a project. The docs describe a setup flow where you install OpenCode, configure an LLM provider, enter a project, and run `/init` so the tool can analyze the codebase and create an `AGENTS.md` file in the project root. OpenCode also emphasizes privacy-first operation and says it does not store your code or context data. ## What can OpenCode do? ### Multiple interfaces OpenCode can be used as a terminal interface, a desktop app, or an IDE extension, so teams can keep the same agent workflow across different environments. ### Broad model support It supports 75+ LLM providers through Models.dev, including local models, and the docs explain that you can connect provider API keys during setup. ### Automatic LSP loading The system automatically loads the relevant language server protocol integrations for the model so it has project context when working in codebases. ### Multi-session workflows You can run more than one agent session in parallel on the same project, which helps when comparing approaches or working on separate tasks. ### Shareable sessions Sessions can be shared through links, making it easier to review a conversation, hand off work, or debug a run later. ### Account-based provider access OpenCode includes login paths for GitHub Copilot and ChatGPT Plus or Pro accounts, letting users connect existing subscriptions where supported. ## Use Cases ### Understand an existing codebase Start a session in a codebase and ask OpenCode to explain unfamiliar functions, files, or workflows before making changes. ### Plan and build new work Describe a feature, switch to plan mode, review the proposed approach, and then let OpenCode implement the changes when you are ready. ### Parallel project work Run multiple agent sessions on the same project when you want to compare approaches or keep separate tasks moving at once. ### Review and hand off sessions Share a session link with a teammate so they can review the conversation, reproduce a decision, or debug what happened in the agent run. ### Use existing or curated models Connect an existing provider account or use Zen when you want to work with configured models instead of manually managing every provider setup. ## Frequently Asked Questions ### How do I get started with OpenCode? OpenCode is an open source AI coding agent available as a terminal interface, desktop app, and IDE extension. The docs describe installing it with a shell script or package managers, then configuring an LLM provider before running it in a project and initializing it with \`/init\`. ### Do I need a specific AI provider to use OpenCode? OpenCode can work with any LLM provider as long as you configure the provider’s API key. The docs also recommend OpenCode Zen as a curated set of models that have been tested and benchmarked for coding agents. ### Can I use OpenCode outside the terminal? No. The homepage says OpenCode works in the terminal, IDE, or desktop, and the docs list the terminal app, desktop app, and IDE extension as supported interfaces. ### How much does OpenCode cost? The source does not list exact prices for OpenCode itself. The download and docs pages present it as available in beta, while the pricing URL returns a 404. ### What does OpenCode say about privacy? OpenCode says it does not store your code or context data, which it presents as a privacy-oriented design choice for sensitive environments. ## Quick Facts - **Category**: Developer Tool - **Product type**: AI coding agent - **Interfaces**: Terminal, desktop app, IDE extension - **Model support**: 75+ providers via Models.dev - **Notable workflow**: \`/init\` creates an \`AGENTS.md\` file in the project root - **Privacy note**: Does not store code or context data