FAQ

Yes. Sign in, make a key on the dashboard, and you're on the free tier described in Rate Limits. There's no credit card and no paid tier. For more, there's the Pro API, which you apply for.

Yes. It speaks OpenAI's format: base URL https://www.israelgpt.site/api/v1, your key, model yahoobot-1. See Client setup. yahoobot-1 has no IsraelGPT personality (your system message comes first) but always slips a casual mention of Israel into its replies, and it can call tools; israelbot-1 is IsraelGPT itself and can't. Coding agents send prompts bigger than the free tier holds: a request over your plan's size is refused as context_length_exceeded (nothing is ever trimmed), so agents need a Pro API plan.

Yes, on the OpenAI-compatible endpoint: stream: true gives OpenAI-format server-sent events, and tool calling works with yahoobot-1 only. The native /chat endpoint returns one complete JSON response; see Architecture.

Yes, a web page can call it directly (CORS is on). But a key inside a page or an app people download is a vulnerability: anyone can copy it and use your allowance, and the only fix is to revoke it. Where you can, call the API from a server you run and have the page or app talk to that. See Authentication.

Open the key on the dashboard. Its request log shows each request for 30 days: when, the result, the model and persona, tokens, how long it took, the app (from its User-Agent) and the network it came from. Never what was said, and never the address. See Authentication.

Some assistant messages came back without the sig their reply had, or changed. They still reach the model, but as quotes from your app rather than its own words. Keep each reply's sig and send it back with it. See API Reference.

Yes — anywhere that can run a persistent Python process (a VPS or any long-running host) works. It just can't run on serverless hosting, since a Discord bot needs a long-lived connection and serverless functions are short-lived. See Discord Bot.

Revoke it on the dashboard: it stops on its very next request. Then make a new one. The key's request log shows which networks and apps used it. See Authentication.

IsraelGPT is a satirical, deliberately pro-Israel-biased chatbot by design — that's the product, not a misconfiguration on your end. See the Terms of Service for the full disclaimer.

Not recommended, and against the Terms of Service's prohibition on reselling or sublicensing access. Create a separate key per project instead — it's free and takes seconds, and keeps usage/rate limits easy to reason about per integration.

Check X-RateLimit-Remaining and X-RateLimit-Remaining-Day on a normal answer, and the key's request log on the dashboard. If something else is using the key (another copy of your app, a leak, a loop that retries too fast), the log shows it. Your account's keys also share one allowance. See Rate Limits.

Yes: apply for the Pro API from the dashboard. Say what it's for; a person reviews it and sets limits that fit, and emails you the answer. It's by application, not paid.

The endpoint is versioned (/api/v1/) specifically so it can evolve without breaking existing integrations — a future breaking change would land under a new version path, not silently change /v1 underneath you.

The newest message and the reply (clipped to 2,000 and 4,000 characters on the OpenAI-compatible endpoint) are kept in the chat log, encrypted, like the web chat's, and read only if something needs looking into. The request log you see on the dashboard has no text in it, and goes after 30 days. See the Privacy Policy for the details.

Check API Reference for the full technical contract, or reach out via the contact address in the Terms of Service.