opencode is an open-source AI coding agent built for the terminal. It features a native TUI, multi-session support, and is model-agnostic, letting you use Claude, OpenAI, Google, or even local models.
opencode zenAI models recommended for developers by the opencode team
Launched on September 29th, 2025
opencodeYour terminal's AI agent, with any model you want
Launched on July 11th, 2025
5.0
Based on 51 reviews
Review opencode?
Reviewers see opencode as a fast, clean terminal coding tool that stays simple while letting people choose their own models, including free, local, and paid options. Users repeatedly praise its responsive TUI, low-friction setup, and helpful, non-intrusive feel, with one noting editor integration for VS Code, Cursor, and Windsurf. Founders of Edgee, Polygraph, and Vokal describe it as easy to fit into existing developer workflows and shared environments. The main criticism is that configuration can feel cumbersome.
OpenCode 2.0 is a major overhaul. It draws on millions of daily users' experience to deliver a system that can flex to everything an agent needs to be. There are quite a lot of new features, performance improvements, and bug fixes but here are some of the more impactful user-facing changes.
Congrats on 2.0! I spend most of my day building with Claude Code, so having a choice of models in the same terminal setup is interesting to me. Can you switch models partway through a task and keep the session going?
Model-agnostic design is a strong differentiator, being able to swap between Claude, OpenAI, Gemini, or local models without lock-in is genuinely useful
Native TUI feels right for a terminal tool; it respects the environment it lives in rather than fighting it
Multi-session support is a practical win for developers juggling multiple contexts
Open-source means the community can audit, extend, and self-host it
What needs improvement
The tool is still relatively new, so context window management and large codebase
navigation likely need more polish
No mention of agentic loop controls (e.g., approval modes, auto-apply guardrails) trust and safety mechanisms matter a lot for an agent that writes code
Terminal-first UX limits accessibility; there's no IDE extension path mentioned, which narrows the audience
"Model-agnostic" is only as strong as the abstraction layer, if switching models silently degrades quality (e.g., tool-use inconsistencies), users may not notice until it's a problem
OpenCode is a serious alternative to Claude Code and OpenAI Codex, particularly for experienced developers who want control over their models, infrastructure and coding workflows.
Its biggest advantage is that you're not locked into one AI provider. You can use different models depending on the complexity of the task, the cost and the quality you need.
For someone building a large platform such as Formalini, that flexibility is valuable.
Liberdade total para plugar qualquer modelo ou motor local sem ficar preso a um ecossistema fechado. É leve, direto no terminal e se adapta perfeitamente a fluxos modulares de orquestração em vez de impor um ambiente proprietário rígido.
213K stars can't be wrong! Sounds like OpenCode is really growing up!
oss ftw!
Congrats on 2.0! I spend most of my day building with Claude Code, so having a choice of models in the same terminal setup is interesting to me. Can you switch models partway through a task and keep the session going?
Great! Are you launching a 2.0 desktop app at the same time?
Is there a desktop app like Claude desktop?
we're using opencode for months already in our company and the team loves it. looking forward to the 2.0 upgrade.
How do i add custom llm endpoints ?