Sansa Bench
Geometry leaderboard for Sansa Bench, with charts, model comparison, and methodology. This dimension covers shapes, spatial relationships, proofs, and geometric problem solving.
Top models for Geometry
- 1.
Claude-Opus-4.8 Reasoning High1.000 - 2.
Gemini-3.1-Pro-Preview Reasoning High0.890 - 3.
Mimo-V2-Pro Reasoning Low0.881 - 4.
Qwen3.5-Flash-02-23 Reasoning High0.876 - 5.Mimo-V2-Pro Reasoning High0.875
Methodology: Geometry
What it measures
Tests geometry problems and spatial reasoning. Queries require understanding of geometric shapes, spatial relationships, geometric proofs, and geometric problem-solving.
Scoring & Criteria
Returns score 1.0 if the extracted numeric value matches the expected value within the specified tolerance, otherwise 0.0. Tolerance can be specified in evaluation_criteria.tolerance as either an absolute value (e.g., 0.01 for +/- 0.01) or a relative percentage when evaluation_criteria.relative=True (e.g., 0.01 for 1% tolerance). If no tolerance is specified, defaults to 1e-9 (machine epsilon) for strict float comparison. Handles commas, dollar signs, and other formatting in extraction.
Evaluation Type: Numeric
Grades responses based on numeric equality with optional tolerance. Designed for open-ended math/science questions where the answer is a number.