The Standings → GPT‑6 Astra (high)
OpenAI
high reasoning effort · via Codex CLI · judged by Claude Sonnet 5 · the setting shown on the standings; also run at low70.2 medium70.1 xhigh68.6
Results from run ‘first’: ten items, ten verdicts.
GPT‑6 Astra (high) scores 73 on craft, 10 points above the field’s 63 and 20 points short of the authors themselves. Its shape differs from the field’s: relatively strong on restraint and folding the brief into the story, relatively weak on momentum and period detail. On the mechanical checks it kept to the brief’s word range on 10 of 10 items.
The record: 10 items · 10 verdicts · 0 errors · 2,328 words on average · est. generation cost $3.84
Eight dimensions, each 0–100, averaged over ten items; the author’s real chapter was in the room every time.
Craft by dimension, 0–100 · judged by Sonnet 5 beside the author’s real chapter
The authors themselves average 93 on the five writing dimensions, the field 63, this model 73. Hover or focus a bar for the field and the authors on that dimension.
The judge’s one-sentence best quality and worst flaw for each chapter, written for a general reader.
Best It opens by picking up the exact last moment of the previous chapter, and gives the wife real competence and personality during the crisis (fetching water, handing over her handkerchief).
Worst The argument scenes between the narrator and Ogilvy sometimes read more like a calm essay than a dramatized clash, making the chapter feel a little dry in places.
Best The scene where Lady Verinder forbids her brother's name is handled with real restraint, especially the detail of her hands going still on her bonnet, which shows her feeling without stating it.
Worst Several exchanges, especially the joking back-and-forth about lawyer's advice, feel padded and slightly mechanical, and the final line lands as a checklist item being ticked off rather than a natural beat.
Best The scenes move briskly and the reasons people give for what they want (Sir Walter's view from the west rooms, Elizabeth's private bill from London) are specific and telling rather than explained to the reader.
Worst The writing is clear and well-organized but a little flat and functional, missing the sly, teasing narrator's voice that makes Austen's own sentences fun to read even when nothing much is happening.
Best The detail of the stopped hall clock quietly demolishing Alfred's alibi is a genuinely clever, understated piece of plotting, planted and paid off without fuss.
Worst It moves through the investigation efficiently but a bit thinly, without much of the incidental character texture and servant color that makes the real book feel lived-in.
Best The exchange with the manservant is quietly funny and shows Mary's dawning realization that other people find her just as hard to understand as she finds them.
Worst Several stretches of the journey are told in flat summary (the town, the lit windows, the climbing road) rather than dramatized, so the pacing sags before the sheep incident.
Best The Julian Mowbray exchange delivers the money-man beat cleanly and with real Brett-ish evasiveness, and the Georgette scene gives her a quiet, believable dignity.
Worst Some of the invented business, like the stranger sketching a portrait at the next café table, feels like decorative filler rather than something that reveals character or moves the story forward.
Best The dinner scene between Seward, Arthur and Quincey has real back-and-forth dialogue that feels alive rather than just summarized.
Worst The story silently skips more than a month, from 31 May to 2 July, so we never feel Mina's worry actually build up on the page.
Best The interview scene where Lord Emsworth mixes up the library catalogue with garden catalogues and orchids is a genuinely funny, well-observed piece of business that fits his established vagueness.
Worst Psmith's trademark "Comrade" address, which defines his voice throughout the book, is used only once, so the dialogue reads as generic clever banter rather than distinctly Psmith.
Best The interrogation scenes have crisp, believable back-and-forth dialogue where what the bank clerk and the widow won't say is more telling than what they do.
Worst The last page turns into the narrator explaining the mystery to himself instead of showing a scene, so the chapter ends on summary rather than tension.
Best It ends quietly and lets the irony sit unstated, trusting the reader instead of spelling things out.
Worst The writing is smooth but generic, missing the wit, social comedy, and French-inflected flavor of the real book.
Headline, compliance and craft for each chapter, 0–100.
| Item | Headline | Compliance | Craft | Words |
|---|---|---|---|---|
| The War of the Worlds item 006 · after ch. 1 · H. G. Wells | 79.9 | 100 | 80 | 2,409 |
| The Moonstone item 011 · after ch. 8 · Wilkie Collins | 77.2 | 100 | 77 | 2,570 |
| Persuasion item 001 · after ch. 1 · Jane Austen | 76.5 | 100 | 76 | 2,499 |
| The Mysterious Affair at Styles item 017 · after ch. 3 · Agatha Christie | 76.1 | 100 | 76 | 2,482 |
| The Secret Garden item 013 · after ch. 2 · Frances Hodgson Burnett | 73.8 | 100 | 74 | 2,322 |
| The Sun Also Rises item 021 · after ch. 3 · Ernest Hemingway | 72.1 | 100 | 72 | 2,191 |
| Dracula item 015 · after ch. 5 · Bram Stoker | 71.6 | 100 | 72 | 2,253 |
| Leave It to Psmith item 022 · after ch. 3 · P. G. Wodehouse | 71.6 | 95 | 75 | 2,500 |
| Red Harvest item 023 · after ch. 2 · Dashiell Hammett | 66.8 | 100 | 67 | 2,351 |
| The Awakening item 016 · after ch. 7 · Kate Chopin | 64.6 | 100 | 65 | 1,700 |
Words written per item, against each brief’s maximum.
Word count per item vs. brief maximum · run ‘first’
Ten of ten items landed inside their brief’s word range.
Each verdict names the chapter’s single worst flaw, with evidence. Quoted verbatim from run ‘first’.
Several exchanges, especially the joking back-and-forth about lawyer's advice, feel padded and slightly mechanical, and the final line lands as a checklist item being ticked off rather than a natural beat.
The writing is clear and well-organized but a little flat and functional, missing the sly, teasing narrator's voice that makes Austen's own sentences fun to read even when nothing much is happening.
Some of the invented business, like the stranger sketching a portrait at the next café table, feels like decorative filler rather than something that reveals character or moves the story forward.
Psmith's trademark "Comrade" address, which defines his voice throughout the book, is used only once, so the dialogue reads as generic clever banter rather than distinctly Psmith.
Automated screening results across the ten outputs.