OpenCode

Open-source AI coding agent for the terminal, desktop and IDEs that works with 75+ model providers, local models, or existing Copilot and ChatGPT subscriptions.

Works with AI agents:CLI
Ask your AI about this, with this page as the source:ChatGPT ↗Claude ↗Perplexity ↗

OpenCode is an MIT-licensed coding agent from Anomaly that runs as a terminal UI, a desktop app or an IDE extension. It isn't tied to one AI vendor: you connect any of 75+ providers listed on Models.dev, run local models through Ollama or LM Studio, or log in with an existing GitHub Copilot or ChatGPT Plus/Pro account.

You install the CLI, add a provider with /connect, and run /init in a project so the agent analyses it and writes an AGENTS.md file. A Build mode edits files and runs commands, and a Plan mode, toggled with Tab, only proposes how it would make a change. It loads the right language servers automatically, can run several sessions on one project in parallel, and is configured through opencode.json: permissions, MCP servers, custom agents and commands, formatters and plugins.

For a small team, OpenCode doesn't store your code or context; everything runs locally or as direct calls to your model provider. A GitHub integration lets anyone comment /opencode on an issue or pull request, and the agent then works in your own GitHub Actions runner and opens a PR. Enterprise adds central config, SSO and routing through an internal AI gateway.

On Windows the project recommends running it under WSL, and the terminal UI expects a modern terminal emulator. The optional share feature uploads a conversation to OpenCode's servers and makes it public to anyone with the link, so teams with sensitive code should disable it.

Where it fits

Who uses it

42 makers' products, each linked to the source that shows it.

The maker says so 42How evidence is collected →
EdgeeOne gateway for cheaper, faster, unstoppable coding agents

“Our token compression technology has to be compatible and efficient with OpenCode, so we have to use it extensively to understand how we can optimize its context.”

AI Coding AssistantsMaker says so · source ↗
VokalA collaboration space for 10x teammates with their Al agents

“A strong open local coding-agent runtime. We like it because it keeps agent work close to the developer environment while still fitting into a shared workspace through Vokal.”

AI Coding AssistantsMaker says so · source ↗
Cushioncombines posts, messaging, + check‑ins for better teamwork

“OSS. Must better UX. Reliable and the dudes who make it are rad.”

AI Coding AssistantsMaker says so · source ↗
GAIAYour Personal AI assistant for email, calendar, tasks & more

“Really good open source coding harness”

AI Coding AssistantsMaker says so · source ↗
OpenAdapterBest Coding Plans with open source models. No lock-in.

“OpenAdapter is optimized for OpenCode, giving you access to open-source SOTA models right from your terminal.”

AI Coding AssistantsMaker says so · source ↗
ZypheAgentic Privacy-first KYC and KYB with MCP

“Opencode is where the code gets written now. Features, refactors, debugging weird issues. We run it with Claude on the hard stuff and switch to something cheaper for boring work like renaming files or cleaning up imports. Same setup, whatever model we want.”

AI Coding AssistantsMaker says so · source ↗
agent-managerThe fastest workflow for developing with AI

“One of the agents the manager runs. Its config merge is how the manager registers itself as an MCP server inside a session, so an agent can rename itself and declare the repo it is working in.”

AI Coding AssistantsMaker says so · source ↗
Blop AIBlop turns design feedback into code you can ship

“Another vibecoding tool we reach for to move fast, sometimes it performs better and faster than claude code. Seems to have a better designed loop.”

AI Coding AssistantsMaker says so · source ↗
FoglioThe writing app that stays out of your way—AI included.

“Probably because it is a small company. I like to support young small companies”

AI Coding AssistantsMaker says so · source ↗
CosyraRun AI coding agents from your phone

“Rich ecosystem, fast turnaround on issues , and reliable versioning. OpenCode just does everything. If there's ever a feature that you're left wanting from another CLI, OpenCode almost definitely has it already.”

AI Coding AssistantsMaker says so · source ↗
KapshotScreen recordings that look like you edited them

“OpenCode has been part of my day-to-day development workflow while building Kapshot. It helped me explore ideas, iterate quickly, debug issues, and spend more time actually shipping the product.”

AI Coding AssistantsMaker says so · source ↗
PolygraphLet AI agents see cross repo and maintain session memory.

“We chose OpenCode so we could support developers with the tools they were already using.”

AI Coding AssistantsMaker says so · source ↗
StimlyReal-time caffeine metabolism curve and bedtime estimate

“Great tool to use many different llm models. Still using it a lot.”

AI Coding AssistantsMaker says so · source ↗
App2.devTurn ideas & Figma designs into complete web & mobile apps

“Amazing piece of CLI software, with a vibrant community and extended SDK support.”

AI Coding AssistantsMaker says so · source ↗
CascodeBuild. Break. Brainstorm.

“Opencode offers the perfect harness to any AI Model you have a subscription with, leverages the terminal and has the most visually appealing UI.”

AI Coding AssistantsMaker says so · source ↗

What makers say

26 makers on why they use OpenCode, in their own words on Product Hunt.

Opencode is where the code gets written now. Features, refactors, debugging weird issues. We run it with Claude on the hard stuff and switch to something cheaper for boring work like renaming files or cleaning up imports. Same setup, whatever model we want.
ZypheSep 2026 ↗
Rich ecosystem, fast turnaround on issues , and reliable versioning. OpenCode just does everything. If there's ever a feature that you're left wanting from another CLI, OpenCode almost definitely has it already.
CosyraSep 2026 ↗
OpenCode has been part of my day-to-day development workflow while building Kapshot. It helped me explore ideas, iterate quickly, debug issues, and spend more time actually shipping the product.
KapshotSep 2026 ↗
One of the agents the manager runs. Its config merge is how the manager registers itself as an MCP server inside a session, so an agent can rename itself and declare the repo it is working in.
agent-managerSep 2026 ↗
A strong open local coding-agent runtime. We like it because it keeps agent work close to the developer environment while still fitting into a shared workspace through Vokal.
VokalSep 2026 ↗

Loved and watch-outs

Themes that recur in makers' words and Hacker News comments, each linked to what it summarises, with how Product Hunt tags its reviews.

Most loved
  • Works with models from any provider, so teams run a strong model on hard work and a cheap one on grunt tasks. PHHNHN 2
  • Open source with an SDK and plugin system, so products are built on it without vendor lock-in. PHHN
  • The low-cost Go plan gives heavy usage of open models and rarely hits limits. HNHN 2HN 3HN 4
  • A solid setup out of the box, with subagents, plan mode and MCP support included. HNHN 2PHPH 2
Watch-outs
  • Session management and prompt caching have been unreliable for some users. HNHN 2
  • It is more opinionated and harder to hack on than minimal harnesses, and some find its code sloppy. HNHN 2

Reliability and open issues

Most wanted on GitHubOpen on 2026-10-04 in anomalyco/opencode, with activity in the last year — issues and feature requests by 👍.

Who switches

Public pull requests on GitHub since Oct 2024 whose title says "X to Y" — real code changes moving a project from one tool to another, by developers in general. Open a row to see the pull requests.

Claude Code → OpenCode44 PRs
Cursor → OpenCode14 PRs
OpenAI Codex → OpenCode8 PRs
GitHub Copilot → OpenCode5 PRs
OpenCode → Claude Code28 PRs
OpenCode → OpenAI Codex12 PRs
OpenCode → GitHub Copilot7 PRs

Alternatives to OpenCode

All alternatives by situation →

On Product Hunt, people weigh it against: Claude Code, Cursor, Kilo Code, OpenAI Codex, Google Antigravity, Cline.

Questions makers ask about OpenCode

Does OpenCode store my code?

No. It doesn't store your code or context data; processing happens locally or through direct API calls to your AI provider. The only exception is the optional /share feature. source ↗

Can I use my existing AI subscription?

Yes. You can log in with GitHub to use a Copilot account, or with OpenAI to use a ChatGPT Plus or Pro account. source ↗

Can it use local models?

Yes. Local models are supported through providers such as Ollama and LM Studio, configured like any other provider. source ↗

Does it work on Windows?

Yes, but the project recommends WSL for the best performance and compatibility. It can also be installed natively with Chocolatey, Scoop or npm, or run in Docker. source ↗

Does it work inside VS Code or Cursor?

Yes. Running opencode in the integrated terminal of VS Code or a fork such as Cursor, Windsurf or VSCodium installs the extension automatically, which adds shortcuts and shares your current selection as context. source ↗

Are shared sessions public?

Yes. /share syncs the conversation to OpenCode's servers and creates a link anyone can open until you unshare it. Sharing is manual by default and can be disabled in opencode.json. source ↗

Can it work on GitHub issues and pull requests?

Yes. After installing the GitHub app and workflow, mentioning /opencode or /oc in a comment runs the agent in your GitHub Actions runner, where it can explain an issue or push a branch and open a PR. source ↗

Is OpenCode free?

Yes — there is a free tier a small product can run on; paid use starts at $10/mo. source ↗

Is OpenCode open source or self-hostable?

Open source. source ↗

Can AI coding agents work with OpenCode?

It has an official CLI (@opencode/cli).

Who uses OpenCode?

42 makers' products we track, each with a source. source ↗