• Compute
  • Customers
  • Pricing
Sign In
More Blog Posts
XDiscordLinkedInYouTube

Products

  • GPUs
  • Inference
  • Studio

Developers

  • Model library
  • Documentation
  • Glossary

Company

  • About Us
  • Blog
  • Events
  • Partnership
  • Scale
  • Career
  • Ambassador program
  • Mission & Vision

Popular models

    Stay in the loop

    By submitting, you acknowledge that we may collect and use the information you provide, which may include personal information.

    XDiscordLinkedInYouTube

    Copyright ©2026 All rights reserved.

    Privacy PolicyTerms of UseLegal Documentation
    More Blog Posts
    other

    OpenCode Now Works with GMI Cloud

    OpenCode is one of the fastest-growing open-source AI coding agents available today, with over 160,000 GitHub stars, 900 contributors, and more than 7.5 million developers/month.

    May 28, 2026

    GMI Cloud is now a native provider inside OpenCode, which means you can connect to the full GMI model catalog directly from the terminal with no custom configuration required.

    This post covers what OpenCode is, what makes it different from other coding tools, and how to get up and running with GMI Cloud in just a few minutes.

    What is OpenCode?

    __wf_reserved_inherit

    OpenCode is an open-source AI coding agent that runs in your terminal.  It is available as a terminal UI, a desktop application, and an IDE extension for VS Code, Cursor, and similar tools.

    Unlike IDE-based tools that require you to switch windows or rely on a single provider, OpenCode treats models as interchangeable. It supports over 75 LLM providers, including Claude, OpenAI, Google Gemini, local models via Ollama, and now GMI Cloud.  You pick the provider, you pick the model, and you can switch mid-session without losing your conversation context.

    OpenCode works directly inside any project folder. It reads your files, reasons about the codebase, and makes changes with your approval. The interface shows you what it is doing at every step, so you stay in control without having to manage individual file edits manually.

    Two Modes Worth Knowing

    OpenCode ships with two primary agents that cover most development workflows.

    Plan is a read-only mode where OpenCode analyzes the codebase and proposes an implementation approach without making any changes.  This is useful before starting a complex feature, when you want to validate assumptions, or when you are working in a sensitive part of the codebase. In Plan mode, you iterate on the approach with the agent until the plan looks right, then switch to Build.

    Build is the default mode with full tool access. The agent reads files, makes edits, creates new files, and runs commands.  You can undo any changes at any point using the /undo command. Switching between modes is done with the Tab key.

    AGENTS.md: Giving OpenCode Context About Your Project

    One of the most useful features in OpenCode is AGENTS.md.  This is a markdown file you place in your project root that contains instructions specific to your codebase: the stack you use, coding conventions, folder structure, testing requirements, and anything else the model should know before touching your code.

    You can generate it automatically by running /init inside OpenCode. The agent scans your project and creates an AGENTS.md with concise project-specific guidance.  This file is included in the model's context at the start of every session, which significantly improves the quality and consistency of responses over time.

    You can also maintain a global AGENTS.md at ~/.config/opencode/AGENTS.md that applies across all your projects.

    Why GMI Cloud

    __wf_reserved_inherit

    GMI Cloud is a venture-backed AI infrastructure company that raised $82 million in Series A funding.  It operates NVIDIA H100 and H200 GPU clusters across data centers in the US and APAC.

    For developers, what matters is the model catalog. Through the GMI Cloud API you get access to Claude, GPT, DeepSeek, Gemini, Qwen3-Coder, and more through a single OpenAI-compatible endpoint.  Because GMI Cloud is a native provider inside OpenCode, you do not need to manually configure base URLs or write custom JSON. You connect once and all available models appear in the OpenCode model picker.

    Connecting GMI Cloud to OpenCode

    Once OpenCode is installed and you are inside a project, type:

    /connect

    A provider list will appear. Scroll down and select GMI Cloud, then press Enter.

    When prompted for an API key:

    1. Go to console.gmicloud.ai
    2. Navigate to API Keys
    3. Click Create API Key and copy it
    4. Paste it into OpenCode and press Enter

    GMI Cloud is now connected. To browse available models, type:

    /models

    The model picker opens showing the full GMI Cloud catalog. You can switch models at any time during a session without losing context.

    Full installation guide here: docs.gmicloud.ai/coding-tools/opencode

    Recommended Models for Coding

    ‍

    Model Best for
    Qwen3-Coder-480B Best raw coding quality, large context tasks
    DeepSeek V4 Pro Complex reasoning and multi-file refactors
    Claude Sonnet 4.6 Balanced quality and speed
    Claude Opus 4.7 Fast, high-quality responses
    DeepSeek V4 Flash Quick edits and lightweight tasks

    ‍

    The right model depends on what you are working on. For complex multi-file tasks or unfamiliar codebases, Qwen3-Coder-480B or DeepSeek V4 Pro give you the most reasoning depth.  For faster iteration on smaller tasks, DeepSeek V4 Flash keeps latency low.

    Useful Commands Inside OpenCode

    ‍

    Command What it does
    /connect Add or switch AI provider
    /models Switch the active model
    /init Generate an AGENTS.md for your project
    /undo Revert the last set of changes
    Tab Toggle between Plan and Build mode
    Ctrl+P Open command palette
    Esc Close any menu

    ‍

    Resources

    • Full installation guide: docs.gmicloud.ai/coding-tools/opencode
    • GMI Cloud console: console.gmicloud.ai
    • GMI Cloud model catalog: docs.gmicloud.ai/model-quickstarts/text/overview
    • GMI Cloud API reference: docs.gmicloud.ai/api-reference/introduction

    Have questions or want to share what you are building? Join the GMI Cloud community on Discord

    ‍

    Roan Weigert

    DevRel AI Engineer

    Build AI Without Limits

    GMI Cloud helps you architect, deploy, optimize, and scale your AI strategies

    Ready to build?

    Explore powerful AI models and launch your project in just a few clicks.

    Get Started