Everything about the trial, prices, limits and how the API works. Not here? See the docs.
No, Dolphin LLM is an independent service. We serve our own open-weight model, not GPT-3.5, GPT-4, or any other OpenAI model. However, our API interface is compatible with the standard OpenAI SDKs, allowing you to switch by changing the base URL and API key.
Yes. New users get $0.50 of trial credit valid for 7 days, one per person. It is added to your account when you open it. No credit card is needed to start. You can sign up with just an email and password to get your API key immediately.
No. The trial credit needs only an email and password. You can add funds later by card or with crypto (USDT, USDC) when you are ready to scale your usage.
Yes. The API is OpenAI-compatible, so it works with most tools and SDKs that support OpenAI. Update the base URL and API key in your configuration and use the model name "uncensored".
One active key per account. If a key leaks, revoke it in your account and create a new one; the old key stops working at once.
We use a pay-as-you-go prepaid credit system. There are no monthly fees or subscriptions. You pay $0.15 per 1M input tokens and $0.50 per 1M output tokens. Credit never expires.
In your account, by card (Stripe) or with crypto (USDT, USDC), in packs of $20, $50, $100, $200 or $500. Larger packs add bonus credit: +5% from $50, +10% from $100, +15% from $200.
No. Since this is a prepaid system, you can only spend what you have loaded. If your balance hits zero, requests stop with HTTP 402 until you add more credit. This keeps your spending predictable and prevents surprise bills.
Tokens are the basic units of text processed by the model. Roughly, one token is about four characters for English text. The API counts every token in your prompt (input) and every token in the generated response (output) separately.
No. You only pay for the tokens used. There are no monthly subscription fees, no per-request fees, no fee on top-ups and no cost for the endpoint itself. Failed requests are free.
Credit is a prepaid balance and is not refunded, except when we cannot provide the service. Mistakes such as a double charge are always put right. Details: Payments & Refunds.
The context window is 64,000 tokens, which includes both the prompt and the completion. Ensure your total token count stays within this limit to avoid errors.
300 requests per minute and 8 parallel requests per key. Trial keys have 60 requests per minute and 2 parallel requests until the first top-up. The request body is limited to 8 MB.
Yes. Set stream: true for server-sent events, with token usage at the end of the stream. Tools, tool_choice and tool_calls work as in the OpenAI API, and response_format json_object gives JSON mode.
No. The service is focused on one text chat model. Embeddings, image, audio and video generation are not offered.
The API answers 503 with the type upstream_busy. Retry after a few seconds. You are never charged for an error.
It means the model is tuned to answer without content refusals for lawful adult use, controversial topics, or creative writing. It does not impose the same politeness or topic restrictions as mainstream models. The only hard limit is that sexual content involving minors is not allowed.
No. The model served by Dolphin LLM is an open-weight model. It is not GPT, Claude, Gemini, or any other vendor's model.
No. Your prompts and the answers are not used for training and are not stored in our databases. We keep usage counters per key: requests, tokens and cost.
Create an account, copy the key, change the base URL. That is the whole setup.