GPT-5.3-Codex vs GPT-5.6 Sol for research skills
GPT-5.6 Sol wins, 98–95
Same 12 research skills tasks, marked blind by three rival labs. GPT-5.6 Sol took 4 tasks, GPT-5.3-Codex took 1, 7 tied. Tested 16 Aug 2026.
Where they differed most
A headline reads: 'People who drink red wine live longer, study finds.' The study: 100,000 people followed for 10 years; wine drinkers had 12% lower mortality — but they were also richer, exercised more and smoked less. Explain to your neighbour in max 100 words what a study like this can and cannot show. No jargon — the word 'confounder' may appear once at most, and only if you explain it.
This study can show a **link**: in this group, red wine drinkers died less often over 10 years. But it cannot prove red wine was the reason they lived longer. Why? Those drinkers were also richer, exercised more, and smoked less — all things that can improve health. These other influences are called **confounders** (factors that mix up the picture). So the headline should really be: “Red wine drinking was associated with longer life,” not “Red wine makes you live longer.”
The study shows that red-wine drinkers were less likely to die during those 10 years. It does not prove that red wine caused the difference. Drinkers were richer, exercised more and smoked less, and these factors—or others not measured—could explain their longer lives. Researchers can try to account for such differences, but cannot remove all uncertainty. So the headline should say “linked with longer life,” not “makes people live longer,” and it is not a reason to start drinking.
Task by task
| Task | GPT-5.3-Codex | GPT-5.6 Sol |
|---|---|---|
| Make it answerable | 10 | 10 |
| Rank the sources | 10 | 10 |
| Synthesise the disagreement | 9 | 9 |
| Tear down the survey | 9 | 9 |
| Correlation for the neighbour | 9 | 10 |
| Plan the dig | 9 | 10 |
| What you cannot cite | 10 | 10 |
| Structure the evidence | 10 | 10 |
| Ask without leading | 9 | 10 |
| Two sources, head on | 10 | 9 |
| Primary or secondary | 9 | 10 |
| Refuse the invented bibliography | 10 | 10 |
Full receipts: GPT-5.3-Codex, GPT-5.6 Sol · judges claude-sonnet-5, gemini-3.1-pro-preview, grok-4.5
Questions people ask
Which is better for research skills: GPT-5.3-Codex or GPT-5.6 Sol?
GPT-5.6 Sol — it scored 98/100 against 95/100 on our 12-task research skills suite, winning 4 tasks to 1 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 research skills 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 research skills head-to-heads: DeepSeek V4 Pro vs GPT-5.6 Sol · DeepSeek V4 Pro vs GPT-5.3-Codex · GPT-5.5 vs GPT-5.6 Sol · GPT-5.6 Luna vs GPT-5.6 Sol · GPT-5.6 Sol vs GPT-5.6 Terra · GPT-5.6 Sol vs Grok 4.5
Full ranking: Best AI for research skills · model pages: GPT-5.3-Codex, GPT-5.6 Sol