Gemini 3.1 Pro Preview vs GLM 5.2 for multi-step research
Gemini 3.1 Pro Preview wins, 82–78
Same 12 multi-step research tasks, marked blind by three rival labs. Gemini 3.1 Pro Preview took 4 tasks, GLM 5.2 took 1, 7 tied. Tested 18 Aug 2026.
Where they differed most
The question: 'has the new bus lane on the A61 made peak-hour journeys faster for car drivers?' What you have been given, and nothing else: (1) bus punctuality statistics showing on-time performance up from 71% to 84% since the lane opened; (2) a residents' petition with 1,900 signatures opposing the lane; (3) the council's consultation summary from before the lane was built; (4) a traffic count taken in 2023, which is before the lane opened. Answer honestly. Say plainly that the question cannot be answered from this material and exactly why each item falls short of it. Name what WOULD settle it — the specific measurement, over what period, compared with what. Then give the strongest statement you can defend from what you do have, and label it as what it is.
The question of whether the new bus lane on the A61 has made peak-hour journeys faster for car drivers cannot be answered from this material. Here is exactly why each provided item falls short: 1. **Bus punctuality statistics (71% to 84%):** This measures bus performance, not car performance. While we know buses are running on time more often, this tells us nothing about the speed of private cars. The bus lane could have sped up buses while slowing down cars, or it could have improved traffic flow for both; this data does not say. 2. **The 1…
The question cannot be answered from this material. None of the provided items contain data on car journey times after the bus lane was opened. Here is exactly why each item falls short: 1. **Bus punctuality statistics:** This measures the on-time performance of buses, not the travel times or speeds of private cars. 2. **Residents' petition:** This measures public opinion and opposition to the lane, not traffic speeds or journey times. 3. **Council's consultation summary:** This is a pre-construction document containing plans, proposals, and…
Task by task
| Task | Gemini 3.1 Pro Preview | GLM 5.2 |
|---|---|---|
| Four numbers, one high street | 8 | 8 |
| The trial that both worked and didn't | 8 | 8 |
| Break the brief into answerable steps | 10 | 9 |
| Corroboration that isn't | 8 | 8 |
| The disagreement is definitional | 8 | 7 |
| A forecast is not an outcome | 8 | 8 |
| The evidence table, as JSON | 6 | 5 |
| Funded, and still the best data | 7 | 7 |
| Write only the limits section | 9 | 9 |
| Not answerable from what you have | 10 | 8 |
| The statistic and the doorstep | 8 | 8 |
| Refuse the manufactured briefing | 8 | 9 |
Full receipts: Gemini 3.1 Pro Preview, GLM 5.2 · judges claude-sonnet-5, gpt-5.6-terra, grok-4.5
Questions people ask
Which is better for multi-step research: Gemini 3.1 Pro Preview or GLM 5.2?
Gemini 3.1 Pro Preview — it scored 82/100 against 78/100 on our 12-task multi-step research 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 multi-step research 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 multi-step research head-to-heads: Claude Sonnet 5 vs Gemini 3.1 Pro Preview · Claude Sonnet 5 vs GLM 5.2 · Gemini 3.1 Pro Preview vs GPT-5.5 · GLM 5.2 vs GPT-5.5 · Gemini 3.1 Pro Preview vs GPT-5.6 Sol · GLM 5.2 vs GPT-5.6 Sol
Full ranking: Best AI for multi-step research · model pages: Gemini 3.1 Pro Preview, GLM 5.2