Nadeemi Beta

Nadeemi Beta: your private AI assistant

Chat in Arabic or English. Pay only for what you use, from prepaid credit. No subscription.

Try Nadeemi free

Get US$0.10 of free credit on this device: enough for a few short replies on the Quick model (Haiku). No email, no phone, no sign-up.

One free trial per device. Free credit is for trying only: it cannot be refunded or withdrawn. The trial uses the Quick model (Haiku), text chat only.

Paste the key you got after your top-up. It stays in this browser only. No key yet? Top up from US$20 to get one.

See a sample answer (free: written in advance, no AI call)

Enter sends. Shift+Enter adds a new line.

Voice

Voice settings

Tip: in voice chat, answers are read aloud and the mic opens again after each answer. Press Esc to stop.

Privacy: voice uses your browser's own speech service. In Chrome your voice is sent to Google to turn it into text; in Edge to Microsoft; in Safari to Apple (some Apple devices do it on the device). Nadeemi never receives or keeps your audio, only the text you send, exactly as when you type. Reading aloud is done by your device; voices marked Online or Natural send the answer's text to the browser maker to make the sound. Voice costs nothing extra.

PDF, images, Word (.docx), Excel (.xlsx), CSV or text. Up to 5 per message, 10 per chat. You can also drop or paste files here.

    Files are read in your browser and sent only with your messages to the AI model (Claude, by Anthropic), to answer you. Nadeemi does not store them: not on our servers, not in our database, not in logs, and not in your saved chats (only the file names). Word and Excel files are turned into text on your device first, so only the text is sent. Anthropic may keep API data for a limited time for safety checks under its own terms. Do not upload passwords, card numbers or ID scans.

    Each file is priced before you send it. A file stays in the chat, so it is sent again with each later message in this chat; prompt caching makes those re-sends much cheaper.

    Estimated cost of this message: –

    Before each message we reserve the most it could cost. You pay only the real usage; the rest is released at once.

    The AI provider remembers the start of this chat for about 5 minutes, so re-sending it costs 5% of the normal input price on Balanced and Deep thinking (10% on Quick). The first message pays a one-off 25% extra on the part it sets up. You pay 1.5x those official prices, like everything else; the receipt line shows what you saved.

    Your chats (saved in this browser only)

    Your chats are kept only on this device, in this browser. They are never sent to or stored on our servers. Anyone who uses this browser can open them, so delete them on a shared computer.

      Prices (US$ per 1 million tokens)

      ModelInputOutputCache read
      Haiku 5.5 (prompts up to 100,000 tokens)
      claude-haiku-5-5      		$0.15$0.75$0.015
      Haiku 5.5, prompts over 100,000 tokens
      claude-haiku-5-5      		$0.75$3.75$0.075
      Sonnet 5.5
      claude-sonnet-5-5      		$3.00$15.00$0.15
      Opus 5.5
      claude-opus-5-5      		$6.00$30.00$0.30
      Fable 5.1
      claude-fable-5-1      		$15.00$75.00$0.375

      Our price is exactly 1.5x the official Anthropic API list price for each model (checked 2026-10-31; lowered from 2x on 31 October 2026). Each reply and receipt shows the exact model ID used. Charges are rounded up to US$0.0001 per message. Thinking tokens count as output.

      Web search costs US$0.015 per search (1.5x the official US$10 per 1,000 searches), at most 2 searches, plus the result text the model reads. Sources are shown under the answer.

      Coming later (not available yet)

      • Coding agent coming
      • Browser and computer tasks coming
      • Nadeemi Books with AI help (a free preview runs in your browser now) coming
      • Email and calendar coming
      • Memory across chats coming