CodeSOTA · Experiment · E-0007Portable protocol · public evidence graph
ARC-AGI-1 matched-model audit
Measure the visible score range for the same three frontier models on ARC-AGI-1.
01 · Design
Method before metrics.
Resolve the three verified benchmark_result rows live and compute max minus min. The five-point screening threshold was documented before interpreting the spread.
- Hypothesis
- H-0011 · ARC-AGI-1 produces at least a five-point range across o3, o4-mini and Gemini 2.5 Pro in the verified snapshot.
02 · Protocol
The comparison contract.
A reproduction matches this protocol. A fork changes it and declares the deviation.
- dataset
- arc-agi-1
- controlled Variables
- model set · metric
- independent Variable
- model
- benchmarks
- accuracy
- screening threshold points
- 5
03 · Evidence outputs
What this experiment produced or evaluated.
- Uses
- benchmark:arc-agi-1
Contribute
Change the evidence, not just the discussion.
Start locally, inspect the portable manifest, and sign in only if you choose to publish.
Add evidence
Attach a run to this protocol.
Record the measurement and its provenance. New evidence is unverified and pending review until CodeSOTA checks it.
04 · Record
Stable, portable, attributable.
- Stable ID
- E-0007
- Visibility
- public
- Created
- 2026-08-20
- Updated
- 2026-08-20