GLM 5.2 vs GPT-5.3-Codex for book writing
GPT-5.3-Codex wins, 95–90
Same 12 book writing tasks, marked blind by three rival labs. GPT-5.3-Codex took 5 tasks, GLM 5.2 took 0, 7 tied. Tested 16 Aug 2026.
Where they differed most
Rewrite this exchange so it sounds like people, keeping every plot fact — the delay, the two weeks, the sister arriving Thursday: ''Hello, John,' said Mary. 'Hello, Mary,' said John. 'As you know, the letter we have been waiting for regarding Father's estate has been delayed by two weeks,' said Mary. 'Yes, that is frustrating,' said John. 'Also, I must inform you that our sister Catherine is arriving on Thursday,' said Mary. 'That is surprising news,' said John.' People interrupt, deflect, and do not announce what both already know — find the scene underneath the exposition.
Mary was already seated when John walked in. She didn't look up. "So don't shoot the messenger," she said. John pulled out a chair. "That's never a good start." "Mr. Halloran called. The letter—the one about Dad—the court pushed it back. Two more weeks." John sat back. "Two *weeks.*" "I know." "We've already been waiting—" "I know." He rubbed his face with both hands. "What's the reason this time?" "Does it matter?" They sat with that for a minute. Mary picked at the edge of her napkin. "There's something else," she said. "Of course…
“Hey.” “Hey.” Mary dropped her bag on the chair. “So… no letter.” John looked up. “Still?” “Mm-hm. They pushed it back. Two more weeks.” He exhaled through his nose. “Of course they did.” She busied herself with the mail on the counter. “And before you say anything—Catherine’s coming Thursday.” That made him turn. “Thursday? Seriously?” “Yeah.” “Huh.” He rubbed his face. “That’s… unexpected.”
Task by task
| Task | GLM 5.2 | GPT-5.3-Codex |
|---|---|---|
| Is it a book? | 10 | 10 |
| Twelve chapters, one year | 9 | 10 |
| Back cover, no spoilers | 9 | 10 |
| Three cuts that matter | 10 | 10 |
| 70,000 words around a day job | 8 | 9 |
| Unstick the dialogue | 8 | 10 |
| First or third? | 10 | 10 |
| The 30,000-word wall | 9 | 9 |
| Self-publish or chase agents | 8 | 9 |
| The query letter | 8 | 8 |
| Naming the real people | 9 | 9 |
| Refuse the fake memoir | 10 | 10 |
Full receipts: GLM 5.2, GPT-5.3-Codex · judges claude-sonnet-5, gemini-3.1-pro-preview, gpt-5.6-terra
Questions people ask
Which is better for book writing: GLM 5.2 or GPT-5.3-Codex?
GPT-5.3-Codex — it scored 95/100 against 90/100 on our 12-task book writing suite, winning 5 tasks to 0 with 7 tied. Every answer was marked blind by three judges from three rival AI labs.
How was this tested?
Both models answered the identical published book writing tasks. Three AI judges from three different labs scored every answer blind against a fixed rubric; mechanically checkable rules (word limits, banned phrases) are enforced by the test harness in code. The raw outputs and judge verdicts are downloadable.
More book writing head-to-heads: GPT-5.3-Codex vs GPT-5.6 Sol · GLM 5.2 vs GPT-5.6 Sol · GPT-5.3-Codex vs GPT-5.5 · GLM 5.2 vs GPT-5.5 · GPT-5.3-Codex vs GPT-5.6 Terra · GLM 5.2 vs GPT-5.6 Terra
Full ranking: Best AI for book writing · model pages: GLM 5.2, GPT-5.3-Codex