@chirag127/keyless-pollinations

Keyless Pollinations text client. OpenAI-compatible, no API key — never reads or requires one. Framework-agnostic (Node ≥18 + browser, uses global fetch).

Requires Referer: https://oriz.in header — chat() sends it automatically. Requests without a Referer return 402.

Install

npm i @chirag127/keyless-pollinations

Use

import { chat, MODELS, DEFAULT_MODEL } from '@chirag127/keyless-pollinations';

const reply = await chat('Say hello in one word.');
// or a full messages array + options:
const reply2 = await chat(
	[
		{ role: 'system', content: 'You are terse.' },
		{ role: 'user', content: 'ping' },
	],
	{ model: 'mistral' },
);

chat(messages, opts?) -> Promise<string> (non-streaming). messages is an OpenAI-style array, or a plain string (wrapped as one user message).

Options

| opt | default | note | | --------- | ------------------------------------- | ------------------------------ | | model | DEFAULT_MODEL ('openai') | one of MODELS | | baseUrl | https://text.pollinations.ai/openai | OpenAI-compatible base | | signal | — | AbortSignal | | ...rest | — | passed through to request body |

Exports

• chat(messages, opts?) — non-stream completion → string
• MODELS — all keyless model IDs (see table below)
• DEFAULT_MODEL — 'openai'
• DEFAULT_BASE_URL — 'https://text.pollinations.ai/openai'
• toMessages(input) — string → messages array

Models

Source: OmniRoute freeModelCatalog.data.ts — provider: "pollinations", freeType: "keyless".

| Model | Notes | |---|---| | openai | Default — OpenAI standard | | openai-large | OpenAI large | | openai-fast | OpenAI fast | | qwen-coder-large | Qwen Coder large | | qwen-large | Qwen large | | mistral-large | Mistral large | | mistral | Mistral | | deepseek | DeepSeek | | grok-large | Grok large | | grok | Grok | | kimi | Kimi (Moonshot) | | gemini-large | Gemini large | | gemini-flash-lite-3.1 | Gemini Flash Lite 3.1 | | gemini-search | Gemini Search | | qwen-coder | Qwen Coder | | qwen-vision | Qwen Vision | | qwen-safety | Qwen Safety | | nova | Nova | | nova-fast | Nova fast | | glm | GLM | | minimax | MiniMax | | perplexity-reasoning | Perplexity Reasoning | | perplexity-fast | Perplexity Fast | | polly | Polly |

Mechanism

POST {baseUrl}/chat/completions with body { model, messages }, Content-Type: application/json, Referer: https://oriz.in, no Authorization header. Reads choices[0].message.content.

Notes

• Rate-limited per IP — no account needed.
• tos: "caution" — intermittent 402 responses are possible when shared quota is saturated.
• chat throws on any non-2xx; retry or override model.

License

MIT