FREQUENTLY ASKED QUESTIONS

Yes — every model, every chat, no subscription tier hiding anywhere. The site is paid for by advertisers: after some answers, a sponsored card appears in the conversation stream. That card is the entire business model. You never see a paywall, a trial countdown, or a pricing page.

No. There is no account system at all — no email form, no password, no phone verification, no credit card. Open the chat page and type your first question. The machine treats everyone as a guest.

Yes — a free, OpenAI-compatible chat completions API at api.adfreellm.com/v1. Click once on the /api page to mint a key (no signup, no email), then point any OpenAI SDK at that base URL. Roughly 500 short requests per day per key. Answers come back with one labeled sponsored block appended — that block is what pays for the call, so please leave it visible to your users. Structured output (JSON mode) and tool-calling requests are never touched.

Five real models, all wired to a streaming backend: Gemini 3.6 Flash (fast, general — the default), DeepSeek V4 Flash (reasoning and code), Gemini 3.5 Flash Lite (fast and light), GLM 5.2 (reasoning, strongest Chinese), and Gemma 4 31B (general, open weights). We never swap models: if the board says DeepSeek, the model answering you is DeepSeek.

Each IP can send about 6 messages per minute. The limit exists because the machines are shared — it keeps responses fast for everyone instead of letting one visitor monopolize the queue. If you hit it, wait a few seconds and send again.

In one place only: after an answer completes, a small sponsored card may appear in the conversation stream, clearly marked as sponsored. No pop-ups, no banners covering the text, no video pre-rolls. The ad sits next to the answer it paid for.

No — and the boundary is enforced by code, not by policy. The model streams its full answer first; only after it finishes does the gateway attach the sponsored card, as a separate data object — never part of the answer text, so copying or exporting an answer never includes the ad. The card is also human-only: on your next message, the gateway strips every ad block from the conversation history before the model receives it, so the AI never reads, references, or imitates an ad. Advertisers pay for placement after the answer — they cannot buy a single word inside it. And for sensitive topics (grief, medical distress, mental health), the system shows no ad at all.

There is no account system, so there is no account data — no profile, no email, no stored identity tied to you. Conversations are handled as guest sessions. Since we never ask who you are, there is nothing personal for us to hold.

The model is simple: as long as sponsored cards cover the compute bill, the terminals stay on. We make no promises about the distant future — but the funding mechanism is visible on every page, which is more transparency than most free services offer.

Use it. Every answered question is what the machine exists for. If a sponsored card is relevant to you, checking it out is what keeps the tokens free. And telling someone who is tired of login screens about AdFreeLLM helps more than anything.