Gemini 2.0 Flash Thinking Experimental (unspecified checkpoint)
Google DeepMind
MathArena reports benchmark results for Google's undated `gemini-2.0-flash-thinking-exp` alias but does not record the resolved dated checkpoint. Because Google separately exposed the `gemini-2.0-flash-thinking-exp-1219` and `gemini-2.0-flash-thinking-exp-01-21` snapshots, this benchmark entry does not assign the alias results to either one; both candidates accept text and images, produce text, and reason intrinsically, but their documented context limits differ.
Benchmark scores
| Benchmark | Score |
|---|---|
| AIME 2025 | 53.3 |
| HMMT Feb 2025 | 35.8 |
| USAMO 2025 | 4.2 |