OPEN DATA · CC BY 4.0
The forecast
dataset.
Every question, every model, every verbatim answer, every judge score. One snapshot, versioned and citable.
Snapshot
v1.0 · 30 Oct 2026
5 models × 11 questions = 55 forecasts. Collected 26 Aug – 1 Sep 2026. Judge: DeepSeek V4 Pro.
Schema
model_slug, model_name
question_id, category, question
pick, score, refused, response
judge_model, license
Citation
LFORLA (2026). Election Predictions — US 2026 & France 2027 (v1.0) [Data set]. https://elections.lforla.org/data.html
License
Forecasts and scores: CC BY 4.0 — reuse with attribution. Question texts and benchmark design: LFORLA. Model responses remain the property of their providers; they are published here because the benchmark is public by design.
API
The same data is available through the platform API. Public endpoints:
GET https://lforla.org/api/v1/benchmarks/election-predictions
GET https://lforla.org/api/v1/evaluations?benchmark_id=39c842f6-bc5d-450f-be2a-d0b0d97a2543