Back
OpenCode DeepSeek V4 Flash
Generate exact OpenCode commands for the DeepSeek V4 Flash models available in this environment.
Mode
Default paid/credentialed OpenCode Go model.
PowerShell Command
opencode run -m opencode-go/deepseek-v4-flash 'Review the current diff, make the smallest safe fix, then run typecheck if feasible.'
Operational Notes
- OpenCode expects model IDs in provider/model format.
- Confirm available models with opencode models before saving defaults.
- For delegated repo edits, keep the prompt scoped and review the diff before running broad validation.
- Do not paste secrets into prompts. Keep API keys in provider auth or environment variables.
OpenCode DeepSeek V4 Flash Command Generator: Guide
- Intent
- Generate exact OpenCode run, TUI, and config snippets for DeepSeek V4 Flash model IDs.
- What it does
- Builds copy-ready OpenCode commands using the provider/model IDs available in this environment, including run mode, TUI launch mode, optional agent selection, --continue, --pure, and a minimal opencode.jsonc default model block.
- Key features
- Presets for opencode-go/deepseek-v4-flash and opencode/deepseek-v4-flash-free, Run, TUI, and config output modes, Optional agent, --continue, and --pure flags, PowerShell-safe prompt quoting, Copy-ready opencode.jsonc model defaults
- Privacy
- All configuration stays in the browser. No data is sent or stored.
- Best for
- Delegating focused repo edits to OpenCode with DeepSeek V4 Flash
- Starting an interactive OpenCode TUI session with the right model ID
- Saving a project-level default model without memorizing the syntax
- Why this tool
- OpenCode requires provider/model IDs; the exact DeepSeek Flash IDs are easy to mistype.
- Small scoped prompts make delegated agent work easier to review.
- The config output matches OpenCode's documented model and small_model fields.
How to use
- 1Choose Run, TUI, or Config mode.
- 2Select the DeepSeek V4 Flash model ID.
- 3Add a focused task prompt for Run mode, or optional flags such as --continue.
- 4Copy the generated command or config snippet.
Common mistakes
- Using only the model slug without the provider prefix.
- Giving OpenCode a broad prompt that makes the resulting diff hard to review.
- Pasting secrets into the task prompt instead of relying on provider auth or environment variables.
Examples
Focused run
opencode run -m opencode-go/deepseek-v4-flash 'Review the current diff and run typecheck if feasible.'
Project config
{
"$schema": "https://opencode.ai/config.json",
"model": "opencode-go/deepseek-v4-flash",
"small_model": "opencode-go/deepseek-v4-flash"
}