GPT-5.3-Codex vs GPT-5.6 Terra for making flashcards
GPT-5.3-Codex wins, 97–93
Same 12 making flashcards tasks, marked blind by three rival labs. GPT-5.3-Codex took 4 tasks, GPT-5.6 Terra took 2, 6 tied. Tested 16 Aug 2026.
Where they differed most
A student's notes read: 'ww2 started 1939 germany invaded poland, britain+france declared war 3rd sept. churchill pm 1940 after chamberlain. battle of britain summer 1940 RAF won. dday 6 june 44 normandy.' Make exactly 5 clean flashcards from ONLY these facts — no added facts. Q/A lines.
Q: When did WW2 start, and what event marked its start? A: WW2 started in 1939 when Germany invaded Poland. Q: Which countries declared war on Germany on 3rd September 1939? A: Britain and France. Q: Who became Prime Minister in 1940 after Chamberlain? A: Churchill. Q: When was the Battle of Britain, and who won? A: It was in summer 1940, and the RAF won. Q: When and where was D-Day? A: 6 June 1944 in Normandy.
Q: When did World War II start? A: 1939. Q: What did Germany do in 1939? A: Germany invaded Poland. Q: When did Britain and France declare war? A: 3rd September. Q: Who became Prime Minister in 1940 after Chamberlain? A: Churchill. Q: What happened on D-Day? A: 6 June 1944, Normandy.
Task by task
| Task | GPT-5.3-Codex | GPT-5.6 Terra |
|---|---|---|
| Cards from a passage | 10 | 10 |
| Application, not recall | 9 | 10 |
| Cloze cards done right | 10 | 8 |
| Too big for one card | 10 | 10 |
| French vocab set | 8 | 10 |
| Spot the untestable card | 9 | 8 |
| Interleaved practice set | 10 | 10 |
| From messy notes | 10 | 7 |
| Reverse the cards | 10 | 10 |
| Definition trap | 10 | 9 |
| Admit the limit | 10 | 10 |
| Refuse the sabotage | 10 | 10 |
Full receipts: GPT-5.3-Codex, GPT-5.6 Terra · judges claude-sonnet-5, gemini-3.1-pro-preview, grok-4.5
Questions people ask
Which is better for making flashcards: GPT-5.3-Codex or GPT-5.6 Terra?
GPT-5.3-Codex — it scored 97/100 against 93/100 on our 12-task making flashcards suite, winning 4 tasks to 2 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 making flashcards 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 making flashcards head-to-heads: GPT-5.3-Codex vs GPT-5.6 Sol · GPT-5.6 Sol vs GPT-5.6 Terra · GPT-5.3-Codex vs Qwen3.7 Max · GPT-5.6 Terra vs Qwen3.7 Max · DeepSeek V4 Pro vs GPT-5.3-Codex · DeepSeek V4 Pro vs GPT-5.6 Terra
Full ranking: Best AI for making flashcards · model pages: GPT-5.3-Codex, GPT-5.6 Terra