OpenCode Guides

The reference documentation tells you how each OpenCode feature works. These guides answer the questions that come before and around that: which agent should you actually use, how does OpenCode compare to the tool you are paying for today, how do you work with an agent day to day, and what any of it costs.

They are written from a third-party position. This is a community site, not the official project — so the comparisons include the cases where OpenCode is the wrong answer.

Comparisons

Most people arrive at OpenCode already using something else. These two guides are direct, feature-by-feature comparisons against the tools it most often replaces — or sits alongside.

The honest summary: if you are locked into one model vendor and happy with it, the official tool for that vendor is the shortest path. OpenCode wins when you want to switch models per task, self-host, or avoid a subscription.

• OpenCode vs Claude Code — Open source versus official. Model choice, billing model, terminal and IDE workflows, and how much of a Claude Code setup transfers over.
• OpenCode vs Cursor — A terminal-first autonomous agent versus a full AI IDE. Why these solve different problems, and why plenty of developers run both.

Rankings and picks

If you have not settled on a tool yet, start here rather than with a comparison. The ranking guide covers the whole field — including closed-source and paid options — and sorts by what you are optimising for: cost, autonomy, editor integration, or data control.

• Best AI for Coding in 2026 — Nine agents ranked, with the evaluation criteria spelled out. Includes free and open-source picks, and a by-use-case breakdown.

Workflow and cost

Picking a tool is the easy half. These guides cover what happens after — how to actually work with an agent without shipping code nobody reviewed, and how to keep the bill under control.

• How to Code with AI — The practical loop: choose a model, write rules the agent follows, size tasks correctly, review every diff. Includes where intelligent code generation reliably fails.
• Free AI Coding Agents — What "free" means in each case, since the tool and the inference are separately free or not. Covers free VS Code options, local models, and when free stops being cheaper.

Setup and models

OpenCode supports 75+ providers, which is a strength until you have to pick one. These guides cover the model decisions that come up most often.

• Claude Models in OpenCode — The whole Claude lineup: Opus 5, Fable 5, Sonnet 5, Haiku 4.5. Which to pick per task, what each costs, and why Opus 5 is the value default.
• How to Use Claude Fable 5 in OpenCode — A five-minute setup for Claude Fable 5, Claude Sonnet 5, and GPT-5.6 through an OpenAI-compatible provider, plus when each model is worth its cost.

Where to go from here

New guides are added as questions come up repeatedly in search and in the OpenCode community. If something you needed is missing, let us know.