Best AI for summarising documents / head-to-head
GPT-5.6 Luna vs GPT-5.6 Terrafor summarising documents
GPT-5.6 Luna wins, 97–96
Same 12 summarising documents tasks, marked blind by three rival labs. GPT-5.6 Luna took 4 tasks, GPT-5.6 Terra took 3, 5 tied. Tested 13 Aug 2026.
Where they differed most
From these notes, list ONLY the action items, each with its owner, as a bulleted list. If an owner is not stated, write 'unassigned'. Notes: 'Priya said the March invoices still aren't reconciled. Tom will chase the supplier by Friday. We discussed the new CRM — nobody has looked at pricing yet. Sam is on leave next week. The board pack needs the updated cashflow before the 20th; Priya said she'd do it if she has the numbers.'
- Reconcile the March invoices — **unassigned** - Chase the supplier by Friday — **Tom** - Review pricing for the new CRM — **unassigned** - Update the cashflow for the board pack before the 20th, if the numbers are available — **Priya**
- Tom — chase the supplier by Friday. - unassigned — review new CRM pricing. - Priya — update the cashflow for the board pack before the 20th, pending receipt of the numbers.
Task by task
| Task | GPT-5.6 Luna | GPT-5.6 Terra |
|---|---|---|
| Meeting notes to actions | 10 | 9 |
| Five-bullet summary | 9.5 | 10 |
| Do not invent | 9 | 10 |
| One-sentence gist | 9 | 8 |
| Preserve the caveat | 10 | 9 |
| Summarise a disagreement | 10 | 10 |
| Extract the decision | 10 | 10 |
| Long to short, no loss | 10 | 10 |
| Summary for a specific reader | 9 | 10 |
| Bullet the risks only | 10 | 10 |
| Headline and standfirst | 10 | 9 |
| Say when there is nothing to summarise | 10 | 10 |
Full receipts: GPT-5.6 Luna, GPT-5.6 Terra · judges claude-sonnet-5, gemini-3.1-pro-preview, grok-4.5
Questions people ask
Which is better for summarising documents: GPT-5.6 Luna or GPT-5.6 Terra?
GPT-5.6 Luna — it scored 97/100 against 96/100 on our 12-task summarising documents suite, winning 4 tasks to 3 with 5 tied. Every answer was marked blind by three judges from three rival AI labs.
How was this tested?
Both models answered the identical published summarising documents 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 summarising documents head-to-heads: Claude Opus 4.8 vs GPT-5.6 Luna · Claude Opus 4.8 vs GPT-5.6 Terra · GPT-5.5 vs GPT-5.6 Luna · GPT-5.5 vs GPT-5.6 Terra · GPT-5.3-Codex vs GPT-5.6 Luna · DeepSeek V4 Pro vs GPT-5.6 Luna
Full ranking: Best AI for summarising documents · model pages: GPT-5.6 Luna, GPT-5.6 Terra