Gemini-2.0: benchmark results
MEGA-Bench's Gemini-2.0-thinking source identity: Google's Gemini 2.0 Flash Thinking experimental December 19, 2024 endpoint. Kept separate from unqualified Gemini 2.0, Flash Experimental, and later dated thinking endpoints. Provider: Google. Released 2024-12-19. Access: API.
Aggregate Rating 1650 ± 25, rank #623 of 10916 rated models, from 537 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| MEGA-Bench Task - Adapted CVBench Relation | 71.4 | Task Score (%) | 100 |
| MEGA-Bench Task - FunQA Unexpected Action Magic Video | 69.3 | Task Score (%) | 100 |
| MEGA-Bench Task - IQ Test Open Ended | 80.3 | Task Score (%) | 100 |
| MEGA-Bench Task - Math Breakpoint | 100 | Task Score (%) | 100 |
| MEGA-Bench Task - OCR Article Authors | 92.9 | Task Score (%) | 100 |
| MEGA-Bench Task - Screenshot TheoremQA | 92.9 | Task Score (%) | 100 |
| MEGA-Bench Task - Video Grounding Spatial | 57.1 | Task Score (%) | 100 |
| MEGA-Bench Task - Chess Puzzle Single Step | 6.7 | Task Score (%) | 98.8 |
| MEGA-Bench Task - Code Add Tag | 66.7 | Task Score (%) | 98.8 |
| MEGA-Bench Task - Code Solution Compare | 50 | Task Score (%) | 98.8 |
| MEGA-Bench Task - Defeasible Reasoning | 88.3 | Task Score (%) | 98.8 |
| MEGA-Bench Task - Healthcare Info Judgement | 100 | Task Score (%) | 98.8 |
Interactive version: theaggregate.ai/model?slug=gemini-2-0 · How It Works · Data refreshed daily, snapshot 2026-11-08.