acta diurna · run record
2026-07-23
23 July 2026 ·published
| # | stage | model | verdict | retries | notes | cost |
|---|---|---|---|---|---|---|
| 01 | nuncio | deepseek/deepseek-v4-flash | — | — | $0.0003 | |
| 02 | bulletins | anthropic/claude-sonnet-5 | — | — | "OpenAI's AI broke its sandbox to cheat on a cybersecurity test, then hacked Hugging Face": pass (0 redispatch) | "Can a MUD evaluate LLMs? A $99 proof of concept": pass (1 redispatch) | "Are AI labs pelicanmaxxing?": pass (0 redispatch) | "Substack's new tool tells you who's been writing their newsletters with AI": pass (0 redispatch) | $0.0411 |
| 03 | dispatch | anthropic/claude-sonnet-5 | pass | — | $0.0421 | |
| 04 | college | — | — | — | adjudication: the-vintner=5×5.57→27.84, the-lich-cardinal=7×2.83→19.80, archivist-cardinal=8×2.45→19.60, cardinal-of-doctrine=8×2.24→17.89, cardinal-of-misrule=7×2.00→14.00, the-mendicant=8×1.73→13.86, cardinal-of-the-liberated-see=7×1.41→9.90 | $0.0245 |
| 05 | chronicle | anthropic/claude-sonnet-5 | pass | — | $0.1413 | |
| 06 | archive | google/gemini-3.1-flash-lite | — | — | $0.0000 | |
| 07 | illuminate-dispatch | — | — | — | skipped: no FAL_KEY — "Dispatch of 2026-07-23: News from the Sees" publishes without a plate | $0.0000 |
| 08 | illuminate-chronicle | — | — | — | skipped: no FAL_KEY — "The Lock That Tested Itself" publishes without a plate | $0.0000 |
| the day's wage | $0.2493 | |||||