Start free. Top up $2 for an hour — credits never expire. Every paid plan ships with Claude chat on your transcripts and the Intelligence Beta summary card.
We email you a one-click login link. Credits land in your account the moment payment clears.
Receipt + login link arrive in your inbox right after payment.
Executive summary, topics & mentions, the key quote — and a chat box where you can ask Claude follow-ups about your own audio. Chat is private and per-session — your messages stay in the tab, nothing is stored on our server, and the next visit starts fresh. The model never trains on your recording.
OpenAI launched three GPT-Realtime audio models on 2026-05-08: Realtime-Whisper streaming STT, Realtime-Translate, and Realtime-2 voice agent. Side-by-side on the dollars, features, and privacy posture — checked 2026-05-08.
| What you pay / get | Whipscribe $2 → 200 min · packs up to $29 / 30,000 min · credits never expire · privacy-first |
ChatGPT Realtime audio OpenAI · launched 2026-05-08 |
|---|---|---|
| Cheapest pack | $2 / 200 min ($0.010/min) credits never expire 30 min / day free |
$0.017 / min ≈ $1.02 / hour (Realtime-Whisper) no free tier |
| Bigger packs (better per-minute) |
$4 / 600 min ($0.007/min) $7 / 2,400 min ($0.003/min) $12 / 6,000 min ($0.002/min) $29 / 30,000 min ($0.001/min) credits never expire |
No packs pure usage metering no caps · no free tier |
| Live translation rate |
$0.010 / minute transcript + your favourite LLM (Claude, DeepL) |
$0.034 / min ≈ $2.04 / hour (Realtime-Translate, 70 → 13 langs) |
| Voice-agent (GPT-5 reasoning) |
Out of scope transcript only; bring your own model |
Realtime-2 $32 / $64 per 1M audio tokens |
| 10 hours / month | $4 Casual pack (600 min · never expires) |
~$10.20 (10 × $1.02/hr) |
| 40 hours / month (power user) |
$7 Creator pack (2,400 min · $0.003/min) ≈ 5.8× cheaper |
~$40.80 (40 × $1.02/hr) |
| 6,000 minutes / month (podcast network) |
$12 / 6,000-min pack ($0.002/min · never expires) ≈ 8.5× cheaper |
~$102.00 (100 × $1.02/hr) |
| Try without paying | 30 min / day anonymous + 120 minutes free on signup |
No free tier billed from minute zero |
| Account / API key required |
Optional paste a URL, get a transcript, no account |
Yes OpenAI account + API key |
| Speaker diarization | Included whisperX + pyannote no extra fee |
Not built-in |
| URL input · batch · library · MCP · GPT |
All five included on every plan |
None developer API only |
| Languages | 99 (full Whisper coverage) |
Whisper-realtime · 99 Translate · 70 → 13 |
| Privacy posture | Self-hosted Whisper audio never sent to OpenAI or any 3rd-party AI no training on uploads |
Audio sent to OpenAI servers (US) retention per OpenAI policy not HIPAA-eligible by default |
Last price-check 2026-05-08. Full side-by-side & FAQ →
Whipscribe is one-time credit packs that never expire — $2 / 200 min ($0.010/min), $4 / 600 min ($0.007/min), $7 / 2,400 min ($0.003/min), $12 / 6,000 min ($0.002/min), $29 / 30,000 min ($0.001/min). At the $12 pack the effective rate is $0.002/minute — about 5× cheaper than ChatGPT Realtime-Whisper ($0.017/min). Realtime is closer at light volumes, but Whipscribe wins past ~5 hours of audio. Whipscribe also includes 30 min/day free without signup, paste-a-URL input, speaker diarization, batch uploads, web library, MCP server, and ChatGPT Custom GPT integration. Audio is processed on Whipscribe’s self-hosted Whisper cluster and never sent to OpenAI. Full comparison →
Yes. Chat runs per-session — your messages live only in the browser tab and are never saved on our server, in our logs, or in your account. Closing the tab clears the conversation. The next time you open the transcript, chat starts fresh. Anthropic does not train on the messages we send through it.
No. Once you buy a pack, the hours sit on your account until you use them. Use them today or a year from now — same price.
No. Every pack is a one-time payment — Razorpay charges you once, the credits land on your account, and they sit there until you use them. There is no monthly bill, no auto-renew, no card-on-file. When the pack runs out you decide whether to buy another one.
Never. We don't use your recordings or transcripts to train, fine-tune, or evaluate any AI model — ours or anyone else's. Transcription runs on Neugence-owned GPUs; nothing is sent to OpenAI, Google, AWS, Deepgram, Rev, or any third-party AI provider.
All prices are in USD and Razorpay charges in USD. If your card is denominated in another currency, your bank converts at its published FX rate and may add a foreign-transaction fee; that's outside Razorpay's control. Indian cards can pay in USD via Razorpay's international-payments flow with no change on your side.
We keep your audio and transcript only as long as needed to deliver the job and let you retrieve it. Ask us to delete sooner at contact@neugence.ai.