gemini

geminiPreset() in tui/internal/preset/preset.go ships provider gemini with the exact default model gemini-3-flash-preview and GEMINI_API_KEY. It uses Google’s native adapter: there is no base_url or OpenAI-compat override. Its manifest includes vision: {provider: gemini, api_key_env: GEMINI_API_KEY}, alongside the DuckDuckGo web_search and default skills capabilities.

Template-specific settings

The model’s inline image input in an ordinary chat turn is native multimodality. The explicit LingTai vision capability is a separate tool path; both use the shipped Gemini provider rather than a fallback provider. Read Google’s official model page and image-understanding guide for current model and image-input details. No official Google plan-level vision MCP is established here. Do not silently switch providers or auto-load/invoke an MCP when the direct route fails. Recheck preset.go for the shipped model, credential env-var, and capability wiring.

Operations

For base URL/API-compat/model/capability declaration shape versus credentials, see reference/operations/endpoint-capabilities/SKILL.md.