NARC puzzles are designed so that neither the grid sequence nor the narrative clue is sufficient on its own — but together they uniquely determine the answer. The experiments below probe different dimensions of this narrative–visual interaction. Masking tests whether models can reconstruct hidden grids. Ordering tests whether narratives help recover temporal structure. Odd-one-out tests whether narratives help models recognize which grids belong together. Stances tests how the same visual pattern responds to different narrative framings.
How to read this page. Masking is the documented battery: every active puzzle, all current models, order and keyword ablations; a paper on it is under review (September 2026). Ordering, Odd-one-out and Stances are pilot designs from spring 2026; each states its own scope and dates below. Counts are per model. The NARC property is a relation between a puzzle and a model, so a model with more NARC puzzles is not a better or worse model, and this page is not a leaderboard.
Ordering experiment: All grids are shown (unmasked) in a shuffled order. The model must recover the correct chronological sequence. Tested under two conditions: grids only and grids + narrative. Agreement with the true order is measured by Kendall's τ (−1 = reversed, 0 = random, +1 = perfect). Narrative lift = τ with narrative minus τ without.
τ = Kendall's tau correlation Lift = τ(grids+narrative) − τ(grids only) Positive lift = narrative helps ordering
Pilot. 4 models, 70 puzzles with four or more grids, data collected on April 3, 2026. Not re-run since and not part of the paper's claims. Report: Can Narratives Help AI Recover the Order of Events? (April 3, 2026, in the code repository).
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | -1.000 | 1.000 | +2.000 |
| gpt-oss-20b | -1.000 | 1.000 | +2.000 |
| nemotron-3-super | -1.000 | 1.000 | +2.000 |
| qwen3.5-122b | -1.000 | 1.000 | +2.000 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | -0.333 | 1.000 | +1.333 |
| gpt-oss-20b | 0.000 | 0.667 | +0.667 |
| nemotron-3-super | -1.000 | 1.000 | +2.000 |
| qwen3.5-122b | -1.000 | 1.000 | +2.000 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | -1.000 | 1.000 | +2.000 |
| gpt-oss-20b | 0.000 | 0.000 | +0.000 |
| nemotron-3-super | -1.000 | 1.000 | +2.000 |
| qwen3.5-122b | -1.000 | 1.000 | +2.000 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | -1.000 | 1.000 | +2.000 |
| gpt-oss-20b | -1.000 | 0.000 | +1.000 |
| nemotron-3-super | -1.000 | 0.000 | +1.000 |
| qwen3.5-122b | -1.000 | 1.000 | +2.000 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | -0.333 | 0.867 | +1.200 |
| gpt-oss-20b | -0.867 | 0.600 | +1.467 |
| nemotron-3-super | -0.467 | 0.867 | +1.334 |
| qwen3.5-122b | -0.867 | 0.867 | +1.734 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | -0.600 | 1.000 | +1.600 |
| gpt-oss-20b | 0.000 | 1.000 | +1.000 |
| nemotron-3-super | -0.600 | 1.000 | +1.600 |
| qwen3.5-122b | -0.200 | 1.000 | +1.200 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | -0.800 | 1.000 | +1.800 |
| gpt-oss-20b | -0.600 | -0.600 | +0.000 |
| nemotron-3-super | -0.600 | 1.000 | +1.600 |
| qwen3.5-122b | -0.800 | 1.000 | +1.800 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | -0.733 | 0.867 | +1.600 |
| gpt-oss-20b | 0.333 | 0.333 | +0.000 |
| nemotron-3-super | -0.867 | 0.867 | +1.734 |
| qwen3.5-122b | -0.733 | 0.333 | +1.066 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | 1.000 | 1.000 | +0.000 |
| gpt-oss-20b | -0.071 | 0.929 | +1.000 |
| nemotron-3-super | -1.000 | -0.071 | +0.929 |
| qwen3.5-122b | -1.000 | 1.000 | +2.000 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | 0.800 | 1.000 | +0.200 |
| gpt-oss-20b | 0.200 | 1.000 | +0.800 |
| nemotron-3-super | 0.200 | 1.000 | +0.800 |
| qwen3.5-122b | -1.000 | 1.000 | +2.000 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | -1.000 | 1.000 | +2.000 |
| gpt-oss-20b | -1.000 | -0.333 | +0.667 |
| nemotron-3-super | -1.000 | -0.333 | +0.667 |
| qwen3.5-122b | 1.000 | 1.000 | +0.000 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | 0.000 | 1.000 | +1.000 |
| gpt-oss-20b | 0.000 | 1.000 | +1.000 |
| nemotron-3-super | 0.000 | 1.000 | +1.000 |
| qwen3.5-122b | 0.667 | 1.000 | +0.333 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | 0.200 | 1.000 | +0.800 |
| gpt-oss-20b | -0.200 | 1.000 | +1.200 |
| nemotron-3-super | -0.200 | 1.000 | +1.200 |
| qwen3.5-122b | 1.000 | 1.000 | +0.000 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | 0.667 | 1.000 | +0.333 |
| gpt-oss-20b | 0.667 | 1.000 | +0.333 |
| nemotron-3-super | 0.667 | 1.000 | +0.333 |
| qwen3.5-122b | -1.000 | 1.000 | +2.000 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | 0.667 | 1.000 | +0.333 |
| gpt-oss-20b | -0.333 | 1.000 | +1.333 |
| nemotron-3-super | 0.667 | 1.000 | +0.333 |
| qwen3.5-122b | 0.000 | 1.000 | +1.000 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | 1.000 | 1.000 | +0.000 |
| gpt-oss-20b | 0.000 | 1.000 | +1.000 |
| nemotron-3-super | 0.000 | 1.000 | +1.000 |
| qwen3.5-122b | 0.333 | 1.000 | +0.667 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | 0.667 | 1.000 | +0.333 |
| gpt-oss-20b | 0.000 | 1.000 | +1.000 |
| nemotron-3-super | 0.000 | 1.000 | +1.000 |
| qwen3.5-122b | 0.667 | 1.000 | +0.333 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | -0.667 | 1.000 | +1.667 |
| gpt-oss-20b | -0.667 | 0.000 | +0.667 |
| nemotron-3-super | 0.000 | 0.333 | +0.333 |
| qwen3.5-122b | -0.667 | -1.000 | -0.333 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | -0.667 | 0.000 | +0.667 |
| gpt-oss-20b | 0.333 | 0.333 | +0.000 |
| nemotron-3-super | 0.000 | 1.000 | +1.000 |
| qwen3.5-122b | 0.000 | 0.667 | +0.667 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | -0.200 | 0.600 | +0.800 |
| gpt-oss-20b | 0.000 | 0.600 | +0.600 |
| nemotron-3-super | -0.200 | 0.600 | +0.800 |
| qwen3.5-122b | 0.600 | 0.600 | +0.000 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | -0.800 | 0.400 | +1.200 |
| gpt-oss-20b | -0.800 | -0.800 | +0.000 |
| nemotron-3-super | -0.800 | 0.200 | +1.000 |
| qwen3.5-122b | 0.400 | 0.200 | -0.200 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | 0.200 | 0.600 | +0.400 |
| gpt-oss-20b | 0.400 | 0.600 | +0.200 |
| nemotron-3-super | 0.600 | 1.000 | +0.400 |
| qwen3.5-122b | 0.000 | 1.000 | +1.000 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | 0.667 | 1.000 | +0.333 |
| gpt-oss-20b | 0.667 | 1.000 | +0.333 |
| nemotron-3-super | -0.667 | 1.000 | +1.667 |
| qwen3.5-122b | 0.333 | 0.000 | -0.333 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | -1.000 | 1.000 | +2.000 |
| gpt-oss-20b | 1.000 | 1.000 | +0.000 |
| nemotron-3-super | 1.000 | 1.000 | +0.000 |
| qwen3.5-122b | 1.000 | 1.000 | +0.000 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | 0.067 | 1.000 | +0.933 |
| gpt-oss-20b | 0.600 | 1.000 | +0.400 |
| nemotron-3-super | 0.733 | 1.000 | +0.267 |
| qwen3.5-122b | 0.733 | 1.000 | +0.267 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | 0.000 | -1.000 | -1.000 |
| gpt-oss-20b | 0.000 | 1.000 | +1.000 |
| nemotron-3-super | -1.000 | 0.000 | +1.000 |
| qwen3.5-122b | 0.333 | 1.000 | +0.667 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | 0.000 | 0.667 | +0.667 |
| gpt-oss-20b | -0.333 | 0.000 | +0.333 |
| nemotron-3-super | -0.333 | -0.333 | +0.000 |
| qwen3.5-122b | 0.000 | 0.667 | +0.667 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | 1.000 | 1.000 | +0.000 |
| gpt-oss-20b | 0.000 | 1.000 | +1.000 |
| nemotron-3-super | -0.400 | 0.000 | +0.400 |
| qwen3.5-122b | 1.000 | 1.000 | +0.000 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | 0.333 | 1.000 | +0.667 |
| gpt-oss-20b | 0.333 | 0.333 | +0.000 |
| nemotron-3-super | 0.333 | 0.333 | +0.000 |
| qwen3.5-122b | 0.333 | 1.000 | +0.667 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | -0.048 | 1.000 | +1.048 |
| gpt-oss-20b | -0.333 | -0.333 | +0.000 |
| nemotron-3-super | 0.524 | 0.619 | +0.095 |
| qwen3.5-122b | 1.000 | 1.000 | +0.000 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | 0.600 | 0.333 | -0.267 |
| gpt-oss-20b | 0.333 | 0.333 | +0.000 |
| nemotron-3-super | 0.333 | 1.000 | +0.667 |
| qwen3.5-122b | 0.333 | 1.000 | +0.667 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | -1.000 | 1.000 | +2.000 |
| gpt-oss-20b | 0.000 | -1.000 | -1.000 |
| nemotron-3-super | 1.000 | 1.000 | +0.000 |
| qwen3.5-122b | 1.000 | 1.000 | +0.000 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | 1.000 | 1.000 | +0.000 |
| gpt-oss-20b | 1.000 | 1.000 | +0.000 |
| nemotron-3-super | 1.000 | 1.000 | +0.000 |
| qwen3.5-122b | 0.000 | 1.000 | +1.000 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | 1.000 | 1.000 | +0.000 |
| gpt-oss-20b | 0.000 | 0.333 | +0.333 |
| nemotron-3-super | 0.333 | 1.000 | +0.667 |
| qwen3.5-122b | 1.000 | 1.000 | +0.000 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | 1.000 | 1.000 | +0.000 |
| gpt-oss-20b | 1.000 | 1.000 | +0.000 |
| nemotron-3-super | 1.000 | -0.200 | -1.200 |
| qwen3.5-122b | -1.000 | 1.000 | +2.000 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | -0.400 | 0.000 | +0.400 |
| gpt-oss-20b | -0.400 | -0.400 | +0.000 |
| nemotron-3-super | -0.400 | 0.000 | +0.400 |
| qwen3.5-122b | 1.000 | 1.000 | +0.000 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | 1.000 | 1.000 | +0.000 |
| gpt-oss-20b | -0.200 | 1.000 | +1.200 |
| nemotron-3-super | 1.000 | 0.600 | -0.400 |
| qwen3.5-122b | 1.000 | 1.000 | +0.000 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | 0.333 | 0.333 | +0.000 |
| gpt-oss-20b | 0.333 | 0.333 | +0.000 |
| nemotron-3-super | 0.333 | 0.333 | +0.000 |
| qwen3.5-122b | -0.667 | 0.000 | +0.667 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | 1.000 | 1.000 | +0.000 |
| gpt-oss-20b | 0.333 | 1.000 | +0.667 |
| nemotron-3-super | 1.000 | 1.000 | +0.000 |
| qwen3.5-122b | 1.000 | 1.000 | +0.000 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | 1.000 | 1.000 | +0.000 |
| gpt-oss-20b | 1.000 | 0.000 | -1.000 |
| nemotron-3-super | -0.667 | 1.000 | +1.667 |
| qwen3.5-122b | 1.000 | 1.000 | +0.000 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | 0.800 | 1.000 | +0.200 |
| gpt-oss-20b | 0.800 | 0.800 | +0.000 |
| nemotron-3-super | 0.800 | 1.000 | +0.200 |
| qwen3.5-122b | 0.800 | 1.000 | +0.200 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | -0.600 | 0.067 | +0.667 |
| gpt-oss-20b | -0.067 | 0.067 | +0.134 |
| nemotron-3-super | 0.000 | -0.600 | -0.600 |
| qwen3.5-122b | -0.333 | 0.067 | +0.400 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | 0.000 | 0.000 | +0.000 |
| nemotron-3-super | 0.000 | 0.000 | +0.000 |
| qwen3.5-122b | -0.333 | 0.000 | +0.333 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | -0.333 | -0.667 | -0.334 |
| gpt-oss-20b | 0.667 | 0.667 | +0.000 |
| nemotron-3-super | 0.667 | 0.667 | +0.000 |
| qwen3.5-122b | -0.333 | 0.333 | +0.666 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | 0.667 | 0.333 | -0.334 |
| gpt-oss-20b | 0.000 | 0.000 | +0.000 |
| nemotron-3-super | 0.000 | 1.000 | +1.000 |
| qwen3.5-122b | 0.000 | -0.333 | -0.333 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | 0.400 | 0.400 | +0.000 |
| gpt-oss-20b | -0.200 | 0.400 | +0.600 |
| nemotron-3-super | 0.600 | 0.400 | -0.200 |
| qwen3.5-122b | 0.600 | 0.400 | -0.200 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | 1.000 | 1.000 | +0.000 |
| gpt-oss-20b | 1.000 | 1.000 | +0.000 |
| nemotron-3-super | 1.000 | 1.000 | +0.000 |
| qwen3.5-122b | 1.000 | 1.000 | +0.000 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | 1.000 | 1.000 | +0.000 |
| gpt-oss-20b | 1.000 | 1.000 | +0.000 |
| nemotron-3-super | 1.000 | 1.000 | +0.000 |
| qwen3.5-122b | 1.000 | 1.000 | +0.000 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | 1.000 | 1.000 | +0.000 |
| gpt-oss-20b | 1.000 | 1.000 | +0.000 |
| nemotron-3-super | 1.000 | 1.000 | +0.000 |
| qwen3.5-122b | 1.000 | 1.000 | +0.000 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | 1.000 | 1.000 | +0.000 |
| gpt-oss-20b | 1.000 | -0.200 | -1.200 |
| nemotron-3-super | -0.200 | 1.000 | +1.200 |
| qwen3.5-122b | 1.000 | 1.000 | +0.000 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | 1.000 | 1.000 | +0.000 |
| gpt-oss-20b | 1.000 | 1.000 | +0.000 |
| nemotron-3-super | 1.000 | 1.000 | +0.000 |
| qwen3.5-122b | 1.000 | 1.000 | +0.000 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | 1.000 | 1.000 | +0.000 |
| gpt-oss-20b | 1.000 | 1.000 | +0.000 |
| nemotron-3-super | 1.000 | 1.000 | +0.000 |
| qwen3.5-122b | 1.000 | 1.000 | +0.000 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | 1.000 | 1.000 | +0.000 |
| gpt-oss-20b | 1.000 | 1.000 | +0.000 |
| nemotron-3-super | 1.000 | 1.000 | +0.000 |
| qwen3.5-122b | 1.000 | 1.000 | +0.000 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | 1.000 | 1.000 | +0.000 |
| gpt-oss-20b | 1.000 | 1.000 | +0.000 |
| nemotron-3-super | 1.000 | 0.667 | -0.333 |
| qwen3.5-122b | 1.000 | 1.000 | +0.000 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | 1.000 | 1.000 | +0.000 |
| gpt-oss-20b | 1.000 | 0.667 | -0.333 |
| nemotron-3-super | 1.000 | 0.667 | -0.333 |
| qwen3.5-122b | 0.667 | 1.000 | +0.333 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | 1.000 | 1.000 | +0.000 |
| gpt-oss-20b | 1.000 | 1.000 | +0.000 |
| nemotron-3-super | 1.000 | 0.667 | -0.333 |
| qwen3.5-122b | 1.000 | 1.000 | +0.000 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | 0.000 | 0.000 | +0.000 |
| gpt-oss-20b | 0.000 | 0.000 | +0.000 |
| nemotron-3-super | 0.000 | 0.000 | +0.000 |
| qwen3.5-122b | 1.000 | 0.667 | -0.333 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | -0.067 | -0.067 | +0.000 |
| gpt-oss-20b | 0.067 | -0.067 | -0.134 |
| nemotron-3-super | 0.067 | -0.067 | -0.134 |
| qwen3.5-122b | 0.333 | -0.067 | -0.400 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | 1.000 | 1.000 | +0.000 |
| gpt-oss-20b | 0.333 | 0.000 | -0.333 |
| nemotron-3-super | 0.333 | 0.000 | -0.333 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | 1.000 | 1.000 | +0.000 |
| gpt-oss-20b | 1.000 | 1.000 | +0.000 |
| nemotron-3-super | 1.000 | 1.000 | +0.000 |
| qwen3.5-122b | 1.000 | 0.000 | -1.000 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | 0.000 | 0.400 | +0.400 |
| gpt-oss-20b | -0.600 | -0.600 | +0.000 |
| nemotron-3-super | -0.200 | -1.000 | -0.800 |
| qwen3.5-122b | -0.200 | -1.000 | -0.800 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | 0.200 | -0.200 | -0.400 |
| gpt-oss-20b | 1.000 | 0.200 | -0.800 |
| nemotron-3-super | 0.200 | -0.200 | -0.400 |
| qwen3.5-122b | 0.600 | 1.000 | +0.400 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | 1.000 | 1.000 | +0.000 |
| gpt-oss-20b | 0.333 | 0.000 | -0.333 |
| nemotron-3-super | 0.333 | -0.667 | -1.000 |
| qwen3.5-122b | 1.000 | 1.000 | +0.000 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | 1.000 | 0.333 | -0.667 |
| gpt-oss-20b | 1.000 | 1.000 | +0.000 |
| nemotron-3-super | 1.000 | 0.333 | -0.667 |
| qwen3.5-122b | 1.000 | 1.000 | +0.000 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | -0.333 | -0.733 | -0.400 |
| gpt-oss-20b | -0.333 | -0.200 | +0.133 |
| nemotron-3-super | -0.333 | -0.733 | -0.400 |
| qwen3.5-122b | 0.000 | -0.733 | -0.733 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | 0.333 | -0.667 | -1.000 |
| gpt-oss-20b | -0.667 | -0.667 | +0.000 |
| nemotron-3-super | -0.667 | -0.667 | +0.000 |
| qwen3.5-122b | 0.000 | -0.667 | -0.667 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | 0.238 | -0.333 | -0.571 |
| gpt-oss-20b | 0.143 | 0.143 | +0.000 |
| nemotron-3-super | 0.238 | -0.333 | -0.571 |
| qwen3.5-122b | 0.238 | -0.333 | -0.571 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | 1.000 | 1.000 | +0.000 |
| gpt-oss-20b | 1.000 | 0.000 | -1.000 |
| nemotron-3-super | 1.000 | 0.400 | -0.600 |
| qwen3.5-122b | 1.000 | 0.400 | -0.600 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | 1.000 | -1.000 | -2.000 |
| gpt-oss-20b | -0.667 | 1.000 | +1.667 |
| nemotron-3-super | -1.000 | -1.000 | +0.000 |
| qwen3.5-122b | 1.000 | -1.000 | -2.000 |
| Model | Grids Only τ | + Narrative τ | Lift |
|---|---|---|---|
| gpt-oss-120b | 1.000 | 1.000 | +0.000 |
| gpt-oss-20b | 1.000 | -0.400 | -1.400 |
| nemotron-3-super | 0.800 | -0.400 | -1.200 |
| qwen3.5-122b | 1.000 | 1.000 | +0.000 |