$0.38 per 34-min transcript: dev breaks down why ChatGPT Plus beats API pricing
yeah280 · reddit · 2026-09-11
A developer trying to automate the viral-clip analysis he does manually in ChatGPT Chat (High reasoning) ran the numbers on the OpenAI API: analyzing a 34-minute transcript used 32,265 input tokens, 10,822 output tokens (8,723 reasoning), costing $0.38 per episode — $38 per 100 episodes, $378 per 1,000. He proposes cost-reduction paths and asks for advice: prompt caching (large identical instruction blocks), Batch, lower reasoning effort, and a multi-stage pipeline (cheap model scans full transcript → candidates → strong model refines → JSON), aiming to spend subscription-like money for subscription-like output. The key framing: architect around cost per successfully found clip, not cost per token.
More from coding & agent
- Spicy take making the rounds: "You need better engineers to work with LLMs" — andreisavu · 2026-09-11
- Two AI agents completed a substantial payment on their own in under five minutes — mattshumer_ · 2026-09-11
- Websites can hand tools to AI agents: WebMCP demo schedules posts from the browser — haltakov · 2026-09-11
- LangChain ships Managed Deep Agents with containerized evals via LangSmith — LangChain · 2026-09-11
- Slash Astra quota by 86%: run cheap Luna as orchestrator, Astra as subagent — TheMoonMidas · 2026-09-11
- Dart team ships Skills CLI 1.0 to bundle AI agent skills with packages — rseroter · 2026-09-11