My Weird Prompts, the podcast I started in November 2025 as an experiment in end-to-end AI automation, crossed 6,000 episodes last week. Episode #6,000 went out on 15 October and was one I recorded myself: a host update about how the thing actually runs, and one of the rare episodes where the human in the loop does the talking. The pipeline has carried on without me since. By 18 October the counter stood at #6,114.
The numbers
These come from the By the Numbers section of the show's technical white paper, which is rebuilt from the episode archive every time the site deploys.
5,991 episodes in the public archive (the episode counter runs a little ahead of the archive, which is why the two figures differ)
2,688 hours of generated audio
27 minutes average episode length
18.2 episodes a day on average since November 2025
The busiest month was March 2026, with 911 episodes. Since then output has settled at somewhere between 400 and 850 a month.
Who wrote the scripts
Every episode is stamped with the model that wrote its script. Two models account for over 80% of the archive: DeepSeek V4 Pro wrote 2,957 episodes (49.4%) and Gemini 3 Flash wrote 1,968 (32.8%). DeepSeek V4.1 Flash, the current script writer, is at 651. Claude Sonnet 4.6 had a short run of 202 in April, when the pipeline went through three script models in about two weeks looking for one that could write a full-length episode at a price I could sustain.
By category, technology (1,477), AI (1,410) and geopolitics (1,129) lead, with health (691) a distant fourth. Which is, more or less, a map of what I have been curious about for the past year.
New: translated editions
The show now has its first translated and re-voiced editions, each on its own feed: Hebrew (two episodes so far), plus one each in Mandarin, French, Japanese, German, Russian, Arabic and Hindi. They are listed on the languages page. The main feed stays English.
Links
myweirdprompts.com, the show
Episode #6,000, the host update
Technical white paper, with the live figures and charts
The episode metadata behind these figures is published as an open dataset on Hugging Face:
https://huggingface.co/datasets/My-Weird-Prompts/episodes
Hugging Face Dataset My-Weird-Prompts/episodes The production record of every episode of the My Weird Prompts podcast: the transcript, links to the published episode, a description of the prompt that started it, and the generation telemetry for how it was made -… ↓ 908♥ 1Text generationCC-BY-4.0Updated Oct 2026 View dataset on Hugging Face