Model	SWE-bench Verified % resolved	GPQA Diamond accuracy	LiveCodeBench Pro Codeforces Elo	Berkeley Function Calling Leaderboard accuracy	LiveBench score	Terminal-Bench 2.1 task success	SWE-bench Pro % resolved	DeepSWE % resolved	Humanity's Last Exam accuracy	MMMU-Pro accuracy	LMArena source-defined rating	ARC-AGI-3 accuracy	ARC-AGI-2 accuracy	FrontierMath accuracy	AIME (OTIS Mock) accuracy	SimpleQA Verified accuracy
Gemini 1.5 Pro Gemini	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—
Gemini 2.0 Flash Gemini	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—
Gemini 2.5 Pro Gemini	—	85.3%	1769	—	—	—	—	—	21.64%	68.0%	1457	—	—	—	—	56.0%
Gemini 2.5 Flash Gemini	—	—	—	56.24%	—	—	—	—	—	—	—	—	—	—	—	—
Gemini 3 Pro Gemini	72.9%	92.6%	2439	72.51%	73.39%	—	43.30%	—	38.3%	81.0%	1479	—	—	37.6%	91.4%	72.9%
Gemini 3 Deep Think Gemini	—	—	3298	—	—	—	—	—	—	—	—	—	84.58%	—	—	—
Gemini 3 Flash Gemini	75.4%	—	2316	—	—	—	34.63%	5.16%	36.6%	—	1466	—	—	35.64%	92.8%	67.4%
Gemini 3.1 Pro Preview Gemini	75.6%	94.1%	2887	—	79.93%	—	46.10%	9.88%	45.9%	80.5%	1481	0.42%	77.08%	36.9%	95.6%	77.3%
Gemini 3.5 Flash Gemini	—	92.8%	—	—	75.02%	—	—	28.32%	42.5%	—	1482	—	72.08%	38.97%	95.6%	68.4%

Scores are not normalised across benchmarks. Each column uses its own metric. Compare columns independently.