Gemini 2.5 Pro

published

Google's frontier reasoning model

Google's thinking model with strong multimodal and coding performance. 89.8% on Global MMLU.

Provider google Type llm Access api Params undisclosed Context 1M License proprietary

Benchmarks (4)

Benchmark	Score	Reported By	Source
MMMU	81.7%	vendor	source ↗
Global MMLU	89.8%	vendor	source ↗
LiveCodeBench v5	70.4%	vendor	source ↗
SWE-bench Verified	63.8%	vendor	source ↗

Best-in-class multimodal reasoning (81.7% MMMU). 1M context window. Very competitive pricing for frontier capability.

API availability varies by region. Gemini models have shown alignment sensitivity issues.

Provider google

Released 2025-03-25

Training cutoff 2025-03

Created March 22, 2026

Last reconciled Never