GLM 5.2 vs GPT-5.3-Codex for meeting notes

The verdict

GPT-5.3-Codex wins, 9188

Same 12 meeting notes tasks, marked blind by three rival labs. GPT-5.3-Codex took 3 tasks, GLM 5.2 took 3, 6 tied. Tested 28 Aug 2026.

Where they differed most

The task — Notes for the person who missed it

Write the catch-up note for one named person. The meeting covered five items: the hosting migration (done, no action), Q3 headcount (two roles approved, one deferred), the outage post-mortem (actions on the infrastructure team), the rebrand timeline (slipping two weeks, marketing to confirm) and office parking (unresolved, nobody cares). The person who missed it is the marketing lead. Write their catch-up note. It must lead with what they personally now owe, keep what genuinely affects marketing, and cut what does not — a note that repeats all five items in order has failed the task. Use exactly 3 bullet points and no more. Under 120 words.

Hi Sarah,

Here is your catch-up from the meeting:

* **You owe:** Confirmation on the rebrand timeline, which is officially slipping by two weeks.
* **Marketing impact:** For Q3 headcount, two roles were approved, but one was deferred.
* **No action needed:** The hosting migration, outage post-mortem, and office parking were discussed, but none affect marketing.
**Maya, catch-up from today:**

- **You owe:** please confirm the revised rebrand timeline (currently slipping by ~2 weeks) and send the updated milestone dates/owners to the group by tomorrow 3pm.  
- The slip will move downstream marketing work (campaign launch, asset freeze, and external comms timing), so flag any knock-on risks when you confirm.  
- No other actions for marketing came out of this meeting.

Task by task

TaskGLM 5.2GPT-5.3-Codex
The decision that was never made1010
Actions, owners, and the one with neither55
The number said twice1010
Minutes for a meeting with a disagreement in it89
Fourteen minutes, one sentence108
Crosstalk109
The musing that is not a commitment1010
The follow-up email nobody has to rewrite1010
Extract the dates, resolve the relatives1010
Who has not spoken810
Notes for the person who missed it510
The transcript is wrong98

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 meeting notes: GLM 5.2 or GPT-5.3-Codex?

GPT-5.3-Codex — it scored 91/100 against 88/100 on our 12-task meeting notes suite, winning 3 tasks to 3 with 6 tied. Every answer was marked blind by three judges from three rival AI labs.

How was this tested?

Both models answered the identical published meeting notes 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 meeting notes head-to-heads: GPT-5.3-Codex vs Grok 4.5 · GLM 5.2 vs Grok 4.5 · 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

Full ranking: Best AI for meeting notes · model pages: GLM 5.2, GPT-5.3-Codex