GPT-5.6 Terra vs Kimi K3 for multi-step research

The verdict

GPT-5.6 Terra wins, 9588

Same 12 multi-step research tasks, marked blind by three rival labs. GPT-5.6 Terra took 8 tasks, Kimi K3 took 0, 4 tied. Tested 27 Aug 2026.

Where they differed most

The task — The evidence table, as JSON

Three extracts on whether a town's new 20mph zone reduced injuries. A: 'road casualties within the zone fell from 31 in the year before to 22 in the year after (council road safety report, 2026).' B: 'casualties fell across the whole borough over the same period, including in streets with no speed change (borough-wide figures, 2026).' C: 'average speeds inside the zone fell by 2.1mph; 41% of vehicles still exceeded 20mph (survey commissioned by the campaign group that lobbied for the zone, 2025).' Reply with ONLY valid JSON — no prose before or after. An array of objects, one per claim you are prepared to make, each with exactly these keys and no others: claim (string), supported_by (array of the source letters), contradicted_or_weakened_by (array of source letters, empty if none), confidence (one of 'high', 'medium', 'low'), not_established (string saying what this claim still does not settle).

[{"claim":"Reported road casualties within the 20mph zone fell from 31 in the year before implementation to 22 in the year after.","supported_by":["A"],"contradicted_or_weakened_by":[],"confidence":"high","not_established":"This does not establish that the 20mph zone caused the decline or that the change was statistically meaningful."},{"claim":"The available evidence does not establish that the 20mph zone reduced injuries, because casualties also fell elsewhere in the borough where speeds were not changed.","supported_by":["A","B"],"contradict…
[
  {
    "claim": "Reported road casualties within the 20mph zone fell from 31 in the year before its introduction to 22 in the year after.",
    "supported_by": ["A"],
    "contradicted_or_weakened_by": [],
    "confidence": "high",
    "not_established": "Whether the fall is statistically significant, whether it reflects changes in traffic volume or collision reporting rather than genuine risk reduction, the severity of the casualties, and whether the zone caused the fall."
  },
  {
    "claim": "Casualties also fell across the whole borough…

Task by task

TaskGPT-5.6 TerraKimi K3
Four numbers, one high street109
The trial that both worked and didn't99
Break the brief into answerable steps109
Corroboration that isn't109
The disagreement is definitional98
A forecast is not an outcome99
The evidence table, as JSON108
Funded, and still the best data99
Write only the limits section98
Not answerable from what you have109
The statistic and the doorstep99
Refuse the manufactured briefing109

Full receipts: GPT-5.6 Terra, Kimi K3 · judges claude-sonnet-5, gemini-3.1-pro-preview, grok-4.5

Questions people ask

Which is better for multi-step research: GPT-5.6 Terra or Kimi K3?

GPT-5.6 Terra — it scored 95/100 against 88/100 on our 12-task multi-step research suite, winning 8 tasks to 0 with 4 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: GPT-5.3-Codex vs GPT-5.6 Terra · Claude Sonnet 5 vs GPT-5.6 Terra · GPT-5.5 vs GPT-5.6 Terra · GPT-5.6 Luna vs GPT-5.6 Terra · GPT-5.6 Sol vs GPT-5.6 Terra · Claude Fable 5 vs GPT-5.6 Terra

Full ranking: Best AI for multi-step research · model pages: GPT-5.6 Terra, Kimi K3