Transcribe audio, video and podcast RSS feeds with Whisper. Get text, SRT/VTT subtitles, Markdown and timestamped RAG chunks from one run. Pay only for speech actually delivered: silent, unreachable or expired recordings cost no audio minutes. No transcription API key.
1.4.3 — 13 September 2026 (not yet built or published)
A run whose every recording was rejected by its own source — silence, an unreachable or
expired URL, a live stream, an unsupported protocol, or a recording past your own duration
or size limit — now finishes successfully and reports each reason as a dataset row, instead
of failing the whole run. Nothing about charging changes: these rows are still 0 audio
minutes. SUMMARY.inputCausedFailureOnly records when this applied.
Any fault on our side — model load, worker exit, transcription or delivery failure, a batch
stop, or uncertain billing or checkpoint state — still fails the run exactly as before.
1.4.2 — 10 September 2026
Three suggested processing choices: Basic/Base, Quality/Small and Max/Medium. Older model: tiny inputs still work; Tiny is not a fourth tier.
Related RAG settings grouped together; machine-readable retry guidance and audio-charge state on errors and run summaries.
Real output screenshots and a short saved-result walkthrough, source-test conditions and explicit scheduled pricing date.
No recognition-model, transcription-quality, resource-budget or tariff change in this patch.
1.4.1 — 10 September 2026
In-flight resource checks, bounded feed/download/SDK operations, lower-overhead decoded sample handling and private recovery checkpoints. Use Resurrect on the original run; reconcile uncertain delivery or billing before retrying. These protections do not guarantee zero loss.
1.3.1 — 10 September 2026
Effective-price preflight and billing receipts, bounded workers, signed file links, timestamp diagnostics, opt-in sentence-aware RAG, paragraphs and updated onboarding.
1.2 — 9 September 2026
Podcast RSS/Atom input and repeat-run checkpoints; Markdown and RAG exports.
1.1 — 6 September 2026
Five-minute audio processing, no-speech failure handling and cached models. The historical 75-minute fixture checked duration acceptance, not representative long-podcast accuracy.
Scheduled pricing — 24 September 2026, 19:51:10 UTC
Basic $0.035, Quality $0.06 and Max $0.12 per rounded-up audio minute. Standard plan discounts apply. Actor-start remains $0.00005 per GB. The existing $0.04 headline minute rate remains effective until then; old pinned builds retain their legacy event. See the Pricing tab for applicable rates.