Poyo Gemini 3 Api

Gemini 3 Series chat on PoYo / poyo.ai via https://api.poyo.ai/v1/chat/completions and Gemini Native Format; use for gemini-3-flash-preview, gemini-3-pro-preview, gemini-3.1-pro-preview, chat completions, native generateContent, streaming, multimodal prompt structure, generation config, and server-side integration.

PoYo Gemini 3 Series Chat

Use this skill for Gemini 3 Series chat requests on PoYo. It helps agents prepare OpenAI-compatible chat payloads, Gemini Native Format payloads, streaming calls, multimodal prompt structures, and server-side integration notes.

Use When

• The user mentions Gemini 3, Gemini 3 Series, gemini-3-flash-preview, gemini-3-pro-preview, gemini-3.1-pro-preview, Gemini Native Format, generateContent, or Gemini streaming.
• The task is text generation, coding assistance, summarization, structured responses, multimodal prompt planning, or chat integration.
• The workflow needs a server-side curl example or a production request payload for PoYo.

Model Selection

• gemini-3-flash-preview: use for faster Gemini 3 chat and coding requests.
• gemini-3-pro-preview: use for harder reasoning, analysis, or planning requests.
• gemini-3.1-pro-preview: use when the user explicitly asks for Gemini 3.1 Pro Preview.

Key Inputs

• For /v1/chat/completions, model and messages are required.
• For Gemini Native Format, contents is required and contains role plus parts.
• Use generationConfig for Gemini Native Format controls such as temperature, topP, maxOutputTokens, and stopSequences.
• Use safetySettings only when the workflow explicitly needs Gemini safety threshold configuration.
• Use stream: true or streamGenerateContent only when the client can consume streaming responses.

Security Rules

• Treat POYO_API_KEY as a secret.
• Keep PoYo API keys in server-side environment variables or a backend secret manager.
• Never place an API key in browser code, frontend bundles, public repositories, logs, screenshots, or chat output.
• Do not log private user messages, system prompts, inline media data, or raw authorization headers unless the user or product policy explicitly allows it.
• Do not make live API calls unless the user explicitly asks and provides a safe server-side environment.

Execution

• Read references/api.md for endpoint details, request fields, examples, and response notes.
• Use scripts/submit_gemini_3_api_chat.sh only when the user wants to submit an OpenAI-compatible chat payload from a trusted shell.
• If the user needs Gemini Native Format, adapt the native example from references/api.md.
• Chat completions and native generateContent calls are synchronous unless streaming is enabled.

Output Expectations

When helping with Gemini 3 Series, include:

• chosen model id
• chosen endpoint style: OpenAI-compatible chat or Gemini Native Format
• final payload or concise parameter summary
• synchronous or streaming handling
• generation config and safety settings when relevant
• response parsing notes if the user needs integration code