Gemini 2.5 Pro

Google DeepMind🇺🇸 United States
active

Google's top reasoning model with thinking mode. Leads on coding benchmarks.

Context window1000K tokens
Input / 1M tokens$1.25
Output / 1M tokens$10

Version History

gemini-2.5-pro-preview-01-14minor

Gemini 2.5 Pro preview January update with longer stable thinking output windows and improved factual grounding on complex queries.

gemini-2.5-pro-preview-08-05minor

Gemini 2.5 Pro preview update with improved tool use accuracy and reduced latency on multi-step reasoning tasks.

gemini-2.5-pro-preview-03-25major

Thinking mode and #1 positions on multiple coding and reasoning leaderboards.

Benchmark Scores

Full leaderboard →
92.0%
AIME 2024
92.0%
AIME 2025
93.1%
DocVQA
88.1%
GPQA
94.3%
HumanEval
77.0%
LiveCodeBench
96.1%
MATH
91.1%
MMLU
82.5%
MMLU-Pro
81.7%
MMMU
62.0 tokens_per_sec
Speed (tok/s)
63.8%
SWE-bench Verified