LegalBench ↗83.8 %
Gemini 2.5 Flash Preview 4/17 (Nonthinking) · LegalBench · Accuracy ↗
Google / MODEL EVIDENCE
Gemini 2.5 Flash Preview 4/17
8 benchmarks · 15 observations · 9 sources
Scores at a glance
8 benchmarks · large dot: Gemini 2.5 Flash Preview 4/17; small dots: closest models where reported
CorpFin v2 ↗54.16 %
Gemini 2.5 Flash Preview 4/17 (Nonthinking) · CorpFin · Accuracy ↗
MortgageTax & TaxEval v2 ↗58.43 %
Gemini 2.5 Flash Preview 4/17 (Nonthinking) · MortgageTax · Accuracy ↗
3 more reported results
71.18 % · Gemini 2.5 Flash Preview 4/17 (Nonthinking) · TaxEval · Accuracy · Vals AI · TaxEval ↗
57.51 % · Gemini 2.5 Flash Preview 4/17 (Thinking) · MortgageTax · Accuracy · Vals AI · MortgageTax ↗
70.52 % · Gemini 2.5 Flash Preview 4/17 (Thinking) · TaxEval · Accuracy · Vals AI · TaxEval ↗
MedQA (USMLE) ↗86.73 %
Gemini 2.5 Flash Preview 4/17 (Nonthinking) · MedQA · Accuracy ↗
1 more reported result
91.02 % · Gemini 2.5 Flash Preview 4/17 (Thinking) · MedQA · Accuracy · Vals AI · MedQA ↗
MATH-500 ↗91.6 %
Gemini 2.5 Flash Preview 4/17 (Nonthinking) · MATH 500 · Accuracy ↗
1 more reported result
91.8 % · Gemini 2.5 Flash Preview 4/17 (Thinking) · MATH 500 · Accuracy · Vals AI · MATH 500 ↗
GPQA Diamond ↗57.58 %
Gemini 2.5 Flash Preview 4/17 (Nonthinking) · GPQA Diamond · Accuracy ↗
MMMU-Pro ↗68 %
Gemini 2.5 Flash Preview 4/17 (Nonthinking) · MMMU Pro · Accuracy ↗
1 more reported result
71.92 % · Gemini 2.5 Flash Preview 4/17 (Thinking) · MMMU Pro · Accuracy · Vals AI · MMMU Pro ↗
LiveCodeBench ↗56.94 %
Gemini 2.5 Flash Preview 4/17 (Nonthinking) · LiveCodeBench · Accuracy ↗
1 more reported result
46.87 % · Gemini 2.5 Flash Preview 4/17 (Thinking) · LiveCodeBench · Accuracy · Vals AI · LiveCodeBench ↗
Closest models by score
Similar positions on shared benchmark leaderboards. Smaller gaps are closer.
DeepSeek V3 →DeepSeek
3 of 3 shared benchmark comparisons
GPT-4o (2024-11-20) →OpenAI
3 of 3 shared benchmark comparisons
Claude Sonnet 4 →Anthropic
3 of 3 shared benchmark comparisons
How similarity is measured
Mean absolute gap in percentile positions on shared benchmark settings; lower is closer. At least 60% overlap is required, with a minimum of three when available. Independent evaluations are used when there are at least three; otherwise published report comparisons are used. Repeated model observations within one setting are averaged before ranking. Writing-style traits and composite indices are excluded. This describes similarity, not an overall quality ranking.
Sources
Every plotted observation links to its evaluation. Reported configurations, benchmark versions and task subsets can differ; similarity describes shared scores, not training lift.
- CorpFin — evaluation methodology and results ↗
- GPQA Diamond — evaluation methodology and results ↗
- LiveCodeBench — evaluation methodology and results ↗
- LegalBench — evaluation methodology and results ↗
- MATH 500 — evaluation methodology and results ↗
- MedQA — evaluation methodology and results ↗
- MMMU Pro — evaluation methodology and results ↗
- MortgageTax — evaluation methodology and results ↗
- TaxEval — evaluation methodology and results ↗
Names used in source tables
Gemini 2.5 Flash Preview 4/17 (Nonthinking) · Gemini 2.5 Flash Preview 4/17 (Thinking)