Questions and answers

One API key for models from Anthropic, OpenAI, Google, xAI, DeepSeek, Mistral and others, plus speech, transcription and image models. It speaks the OpenAI and Anthropic API formats, so existing SDKs and tools work by changing the base URL and key. We also make the SparrowCodes CLI, a coding agent for your terminal.

No. Every account starts on the free plan: 50 requests a day with the free model, and free text-to-speech characters every day. Add credits or buy a plan when you need more.

Set the base URL to https://api.sparrowcodes.com/v1 (or the same address without /v1 for Anthropic SDKs), use your SparrowCodes key, and use a model id such as anthropic/claude-sonnet-5.5. Anthropic's own model ids work on /v1/messages too.

Claude Code, Cline, Continue, Aider and Cursor, and the OpenAI, Anthropic, Vercel AI, LangChain and LiteLLM SDKs. Anything that lets you set an OpenAI or Anthropic base URL should work.

Each model has a price per million tokens, shown on the models page. Anthropic, OpenAI and Google models cost what those providers charge. When you add credits there is a 5.5% platform fee, lower for large top-ups.

Credits are pay as you go: every request is charged at the model's price. A coding plan, from $10, gives you a fixed amount of usage for 30 days for one price, with limits every 5 hours and every week so one busy day doesn't use up the month.

No. Plans last 30 days and never renew by themselves. Buying the same plan again adds another 30 days of usage on top of what's left.

PayPal, card or UPI. Prices are shown in US dollars or Indian rupees. Customers in India pay GST and get a tax invoice for every payment.

Unused credits can be refunded if you ask within 7 days of the payment. See the refunds page for the details.

It can't. We reserve a request's worst-case cost before it starts and release what isn't used, so a request is refused up front rather than cut off halfway.

A coding agent for your terminal on macOS, Linux and Windows. It reads your project, edits files and runs commands and tests. It asks before it runs a command or changes a file.

It picks a model for each task: fast, low-cost models for routine work and stronger ones for hard work, moving up when a task keeps failing. It stays within your plan. You can always pick a model yourself.

Yes. The free plan gives 50 requests a day on the free model.

It sends the parts of your project it reads to the model you're using, through SparrowCodes, like any coding agent. We don't store prompts or answers.

No. We don't store chat prompts or answers. Usage records keep the model, token counts, cost and timing. The response cache only stores answers when you turn it on.

That depends on each provider's terms. Add "provider": {"data_collection": "deny"} to skip providers whose terms allow training on API data. Free models may keep request data, so don't send secrets to them.

Keys start with sk-sc-v1- so secret scanners recognise them. If GitHub finds one in public code, we disable it within seconds and email you. You can also set a spend limit on every key.

Every account gets free characters every day, shared across the speech models. After that you pay per million characters. The speech page shows the current allowance and price.

Yes, with Pocket TTS. Record or upload 10 to 30 seconds of the voice in the dashboard, then use it like any other voice.

Recraft V4.1 Flash, FLUX 3 and FLUX 3 Klein, Seedream 5.0 Flash, Qwen Image 3, Nano Banana 2.1, Krea 2 and GPT Image, all through /v1/images/generations and priced per image.